PART II
FRBR AND OTHER SOLUTIONS
©Karen Coyle CC-BY
©Karen Coyle CC-BY
F OUR
FRBR IN
CONTEXT
N
early twenty years after the first draft of the Functional Requirements for
Bibliographic Records (FRBR), and fifteen years after its final draft was
released, the use of concepts and structures defined in FRBR is assumed
to be the way forward. Yet it isn’t at all clear to me that we have an understanding
of what FRBR means for library practice and for library users.
The FRBR model of bibliographic data is the most radical change to library
catalog thinking since Panizzi developed his ninety-one rules for the British
Museum Catalog. The model presented by FRBR is complex, but it is made even
more complex by the competing concepts in the Final Report from the FRBR
Study Group that developed FRBR (IFLA 2009). FRBR is alternately seen as
an analysis of user needs, a description of the cataloging workflow, and as a data
model for a future bibliographic record format. It is rarely viewed, however, as
what it was originally intended to be: as the development of basic requirements
for an international standard national bibliographic record.
65
©Karen Coyle CC-BY
66 / CHAPTER FOUR
Few of us have done a close reading of the Functional Requirements for Bibliographic Records document, although I would guess that many have glanced at
the diagrams, either within the context of the document or as illustrations used
in presentations that we’ve attended. This means that many of us have had our
concept of FRBR formed from secondary sources that emphasized only a portion
of the content of the document. The FRBR Final Report is 142 pages in length,
including appendixes, which makes it a formidable read. The document includes
three very high-level entity-relation diagrams—high-level in the sense that they
contain very little detail. Although a picture may be worth a thousand words,
these three diagrams are far from expressing the full meaning of the work of the
FRBR Study Group. There is some ambiguity between the textual description
of the conceptual model and the entity-relation diagrams that have come to
represent FRBR for most librarians.
Most discussions of FRBR begin with a list of the entities in the three groups,
and then illustrate these entities with one or more of the diagrams from the document. I hope to do something quite different here, which is to focus on the text
itself, and how the text describes goals and conclusions of the study. I also will
provide some historical context for the work of the Study Group.
With the implementation of FRBR concepts in the cataloging standard Resource
Description and Access (RDA), FRBR has been accepted by some members of
the library community. However, there have been few studies testing the library
user’s view of FRBR and there are not a great number of implementations of
FRBR as an actual catalog.
A report produced in 2006 for the Library of Congress by Karen Calhoun
recommended investigating FRBR, which showed that the author did not consider
FRBR a “given,” but only one possible direction for bibliographic data: “4.2.6
Support experimentation with FRBR and urge vendors and library service organizations to implement clustering based on FRBR concepts” (Calhoun 2006, 18).
The 2008 Library of Congress report on the Future of Bibliographic Control,
titled On the Record recommended that all work on RDA be halted while studies
can be done on the viability of FRBR. Although these two major LC reports
called for systematic investigation of the ideas presented in FRBR, that did not
happen In the meanwhile, FRBR concepts were incorporated into RDA, which
had an implementation date of March 31, 2013, for participating libraries.
The analysis in this book takes a broad view of the cataloging culture that
preceded FRBR in an attempt to understand the motivation of the members of
the FRBR Study Group. What problems were they trying to solve, and what
were the tools at their disposal?
©Karen Coyle CC-BY
FRBR IN CONTEXT / 67
We will see that the primary direction taken by FRBR Study Group, using
an entity-relation analysis model, greatly influenced the outcome of the study.
Proponents of that method in the FRBR Study Group were also key members of
the cataloging standards community that developed the successor to AACR that
began shortly after the publication of the FRBR Final Report. After a false start
on AACR3, the cataloging rules were reborn as rules for an implementation of
the FRBR entity-relation model, RDA.
It is clear that these two standards, FRBR and RDA, were heavily influenced
by the thinking of a small group of people, perhaps no more than a score. Even
if the meetings of the FRBR Study Group were open to the public, the standard
was developed by a group with a closed membership and who did not use available
social media to extend the conversation and deliberation beyond itself. Comments
were solicited from IFLA institutional members, but, of course commenting on
a draft of a document is far from participating in its creation. That said, this is a
common standards development method in the library community, which contrasts
with the groups that develop the Internet and the World Wide Web. The Internet
Engineering Task Force allows anyone to make proposals relating to the technology. Those proposals are called (and remain throughout their use) “Requests
for Comments.” Changes are not only discussed; they are implemented in code
as the proposal progresses through discussion. Anyone can participate in the
development of standards. The World Wide Web Consortium, another standards
development body, does have members—over 400, in fact. Members are companies and institutions. Some committees are limited to member representatives,
but most communication takes place on open mailing lists to which anyone can
post, and document drafts often are developed on publicly accessible wiki pages.
Members can submit documents that discuss or propose web-related technology.
Another significant difference between these library standards and standards
in other communities is that library standards not only do not provide proof of
concept through “running code,” they actually eschew technology altogether.
At least, they claim to. Both FRBR and RDA are stated to be “technology neutral.” This is obviously not true, because the analysis in FRBR made use of a very
specific technology, the entity-relation model. Perhaps it would have been more
accurate to say that FRBR was “application neutral.” However, it is probable
that the members of the group did not understand how much the technology of
relational model determined the group’s outcome. As we’ll see, there is at least
some evidence that the entity-relation model was not well understood, and that
this has resulted in some contradictions between statements in the text of the
report and presentation of the model as entity-relation diagrams.
©Karen Coyle CC-BY
68 / CHAPTER FOUR
What is notable about FRBR, and in some respects RDA also, is that it makes
numerous assumptions that were never tested. Because FRBR was couched in
terms of a known technology, it was assumed to be technically valid and perhaps
even implementable, in spite of the declarations of technology neutrality. Yet no
implementations of FRBR, even on a small set of test data, were developed as
part of the FRBR Study Group’s process. RDA is therefore a cataloging standard
based on an unproven conceptual model. The technology that would support
them is, at the time of this writing, still unavailable.
In spite of lack of proof of FRBR as a bibliographic model, the concept of
FRBR has reached beyond the library community. These implementations often
differ considerably from the presumed library implementation. Unfortunately,
these variations generally do not provide an explicit statement of their interpretations of FRBR or why they chose a different reinterpretion of FRBR as defined
in the FRBR Final Report.
To understand how the library community got to this point, it is necessary to
revisit the context in which the FRBR standard was developed.
©Karen Coyle CC-BY
F IV E
FRBR: STANDARD FOR
INTERNATIONAL SHARING
FRBR
STANDARD FOR
INTERNATIONAL
SHARING
A
s I write this in 2015, the term FRBR has meaning on its own. The
full form, Functional Requirements for Bibliographic Records, is still
known but has faded into the background. The concept that that library
world needed functional requirements for its bibliographic records was foremost
in 1992, when the Study Group on Functional Requirements for Bibliographic
Records was formed as an IFLA group. We must go back to that time to understand the original intention of the task, and what problems it was asked to solve.
The impetus to define functional requirements for bibliographic records (FRBR)
arose out of the IFLA-sponsored Stockholm Seminar on Cataloguing in 1990. The
IFLA cataloging section had been addressing international cataloging standards
for decades, most prominently through the creation of the International Standards
for Bibliographic Description (ISBD) and the International Cataloguing Principles
(ICP1961). I don’t find evidence that the seminar produced any of its own documents; 1990 predated the IFLA web site, which is where outcomes are reported
today. I therefore rely on reports written at later dates. Significantly, the key reports
69
©Karen Coyle CC-BY
70 / CHAPTER FIVE
were written by American members of the FRBR Study Group, and therefore may
not reflect exactly the point of view of other participants at the Stockholm Seminar.
Barbara Tillett’s 1994 report on the meeting and its outcomes describes the
factors that led to the assignment of the FRBR Study Group; she refers to “the
mounting costs of cataloging,” the proliferation of new media, “exploding bibliographic universe,” the need to economize in cataloging, and, “the continuing
pressures to adapt cataloguing practices and codes to the machine environment.”
Regarding the concerns about the costs of cataloging, she states: “Some speakers proposed that cataloguing could be considerably simplified. One speaker
stated that the number of descriptive data elements needed in a bibliographic
record could be reduced without seriously affecting access” (Tillett 1994). In
that same document, Tillett states that the members of the Seminar could not
reach consensus on the “functions of bibliographic records,” and failed to reach
“common agreement on what the bibliographic record is to achieve in answering
user needs.” This is a strong statement about the perceived state of cataloging
in 1990, 150 years after Panizzi drew up his ninety-one rules, 110 years after
Cutter’s statement of the objectives of the catalog, fifty years after the first International Standard for Bibliographic Description, and over two decades after the
creation of the detailed cataloging rules in the Anglo-American community.Yet
this harks back to Seymour Lubetzky’s 1946 criticism of the cataloging rules: they
appeared to be arbitrary because they did not include a functional justification
for the purpose of each rule. For catalogers, it might have been disconcerting
to discover that the rules that they had been applying for years did not have any
specific user goals behind them.
Olivia Madison, who chaired the FRBR Study Group for part of its time,
reported on the group’s results at a meeting of the International Congress on
National Bibliographic Services in Copenhagen in 1998. Her summary of the
activity stated: “The central goals of this study were to assist in decreasing the
costs of cataloguing by encouraging the sharing of bibliographic data records
and to recommend the most useful and important data elements in those records
for their users” (Madison 1998).
The goals, then, that prompted the formation of the FRBR Study Group were:
``
``
``
Determine a minimum set of data elements needed to satisfy user needs.
Reduce the costs of cataloging.
Encourage sharing of bibliographic records (internationally).
It is interesting to compare these to the goals of ISBD. The ISBD document
opens with this statement of its goals:
©Karen Coyle CC-BY
FRBR: STANDARD FOR INTERNATIONAL SHARING / 71
The primary purpose of the ISBD is to provide the stipulations for compatible
descriptive cataloguing worldwide in order to aid the international exchange of
bibliographic records between national bibliographic agencies and throughout
the international library and information community. (ISBD 2011, 1)
One would think that “basic bibliographic data” and “consistency when sharing”
would fit in with the goals that came out of the Stockholm Seminar. Given the
proximity of these goals to those of the FRBR study, it is not clear why ISBD,
or an ISBD variant, was not considered as an answer to the needs as stated, particularly because of the emphasis on “bibliographic records,” which is precisely
the area in which ISBD performs (and FRBR does not). Comparisons of ISBD
and FRBR show how close they are in terms of data elements, which appears
to be by design, not by accident. The Final Report of the FRBR Study Group
hints at this when it says:
The attributes defined for the study were derived from a logical analysis of the
data that are typically reflected in bibliographic records. (FRBR Final Report, 31)
In terms of defining a “minimum set” of bibliographic elements, by my count
ISBD has a few less than one hundred data elements, while FRBR has about
eighty-five, which doesn’t reduce the number of elements in any significant way.
Some minor adjustments were made to ISBD to avoid conflicts with FRBR, but
any other interrelation is difficult to describe. As we’ll see, FRBR does not include
cataloging instructions nor display information, although both are important
aspects of ISBD. In spite of the fact that the term bibliographic records is in the
name of the standard, FRBR does not address record structure, while ISBD does,
although not as the term record would be interpreted by a programmer or database
designer. ISBD adherence should result in relatively consistent textual output
from diverse cataloging departments, and the resulting data, with its specialized
punctuation, should be comprehensible even when the language of cataloging
is not understood. Because FRBR does not define a record format, there is no
equivalent in FRBR to the ISBD punctuation and display rules.
THE FRBR TERMS OF REFERENCE
Coming out of the Stockholm Seminar, a strong motivation for the development
of a new bibliographic model was economics: the need to serve users while reducing the cost of cataloging worldwide. The economic issue is addressed repeatedly
©Karen Coyle CC-BY
72 / CHAPTER FIVE
in the introductory section of the Final Report of the FRBR Study Group, with
statements such as:
The purpose of formulating recommendations for a basic level national bibliographic record was to address the need identified at the Stockholm Seminar for
a core level standard that would allow national bibliographic agencies to reduce
their cataloguing costs through the creation, as necessary, of less-than-full-level
records, but at the same time ensure that all records produced by national bibliographic agencies met essential user needs. (FRBR Final Report, 2)
A document called the Terms of Reference, authored by Tom Delsey and Henriette Avram, gave the official charge to the FRBR Study Group. It defined the
problem and the intended outcomes of the group’s work.
As defined in the Terms, the FRBR Study Group would consist of group
members as well as consultants chosen to draft the report. The first consultants
were Elaine Svenonius, Barbara Tillett, and Ben Tucker. Over the course of the
work, the group of consultants changed. Tom Delsey joined the group a year
later when Tucker left. Later, Elizabeth Dulabahn of the Library of Congress
was added.
The consultants had four required tasks, as laid out in the Terms of Reference:
1.Determine the full range of functions of the bibliographic record and then
state the primary uses of the record as a whole.
2.Develop a framework that identifies and clearly defines the full range of entities
(e.g., work, texts, subjects, editions, and authors) that are the subject of interest
to users of a bibliographic record and the types of relationships (e.g., part/
whole, derivative, and chronological) that may exist between those entities.
3.For each of the entities in the framework, identify and define the functions
(e.g., to describe, to identify, to differentiate, to relate) that the bibliographic
record is expected to perform.
4.Identify the key attributes (e.g., title, date, and size) of each entity or relationship that are required for each specific function to be performed. Attribute
requirements should relate specifically to the media or format of the bibliographic item where applicable.
The final requirement was assigned to the FRBR Study Group itself:
5.For the National Libraries: for bibliographic records created by the national
bibliographic agencies, recommend a basic level of functionality that relates
©Karen Coyle CC-BY
FRBR: STANDARD FOR INTERNATIONAL SHARING / 73
specifically to the entities identified in the framework and the functions that
are relevant to each.
The Terms are more specific than the goals that came out of the Stockholm
Seminar; they clearly define how the Group is to go about its work, by defining entities and relationships, a modeling method that had been developed for
relational database design. The Terms even include a reference to Data Analysis:
The Key to Data Base Design by Richard C. Perkinson (1984), as background for
performing an entity-relation analysis. According to Olivia Madison, who was chair
of the FRBR Study Group from 1991 to 1993 and again from 1995 to 1997:
As mentioned earlier, Delsey had originally authored, with Avram, the initial CDNL
[nb: Conference of Directors of National Libraries] Terms of Reference, and was
well versed in the research literature and potential applications of E-R modeling.
In fact, it was largely due to Delsey’s commitment to this modeling technique
that it was explicitly included in the CDNL Terms of Reference. (Madison 2005)
The Terms of Reference were accepted at the 1992 IFLA Conference in New
Delhi, after two additions were made by the Standing Committee of the Section
on Cataloguing (Madison 2005):
1. Subjects were added to the list of entities.
2.The FRBR Study Group was charged with proposing minimal level standards for bibliographic records.
It is particularly interesting that the first version of the Terms of Reference ignored
both subject access (which surely has user implications) as well as the primary
purpose of the study, which was to address the costs of cataloging for national
bibliographic agencies by creating a minimal-level record requirement. The version
of the Terms of Reference published in 1992 makes only a brief mention of the
economic goals, referring to “operating under increasing budgetary constraints
and increasing pressures to reduce cataloging costs through minimal-level cataloging.” The remainder of the Terms document focuses on a technical analysis
of bibliographic data.
In the end, what came to dominate the outcome of the FRBR Study Group’s
work had little to say about minimal level records or about addressing economic
issues. The “user tasks” that have become a near mantra in some cataloging circles
(“find, identify, select, obtain”) are placed adjacent to bibliographic elements in
one section of the FRBR Final Report, but their relation to the user is left unstated.
©Karen Coyle CC-BY
74 / CHAPTER FIVE
During the development of FRBR there were periods of review, and comments
were received, but these have not been made public. (For a profession that prides
itself on preserving knowledge for the future, we are terrible at keeping our own
history.) Olivia Madison recounts a 1996 review of FRBR that received forty
responses from sixteen different countries. Of these, seven were critical of FRBR.
The principal issues reflected honest difference of professional perspectives
related to the research methodology, the actual need for this particular study,
the adequacy of its user-focus, and the work process and adequacy of detail.
(Madison 2005)
It would be interesting to know which libraries responded, and what those differences of opinion were. That criticism was made of the research methodology
(which may or may not refer to the insertion of an E-R analysis into the study)
and the adequacy of its user focus seems important from today’s perspective.
Madison calls these forty responses from sixteen countries a “worldwide review,”
although it seems a bit “world-narrow.” In the end, FRBR was developed and
reviewed by a very small constituency that was not representative, by any measure,
of the library community that prompted its development. That said, the use of
select expert committees to develop standards is not unusual, when few libraries
can give staff leave to devote themselves to such demanding activities. While this
is understandable, one must acknowledge the fact that the end result of such a
process may not actually serve the larger library community.
©Karen Coyle CC-BY
S IX
T H E E N T I T YR E L AT I O N M O D E L
N
early every book, article, or presentation about FRBR has an explication
of the primary FRBR entities and their relationships. The entity-relation
(E-R) analysis defines the primary structure of the FRBR Final Report,
which has chapters for entities, relationships, and attributes. The use of the entity-relation modeling technique was a requirement posed by the Terms of Reference
for the study. Because of the great influence that this modeling technique had on
the outcome of the study, it is worth examining in some detail.
Entity-relation analysis makes use of particular notation or diagrams to explain
what is being expressed. There are many possible notation styles, some using boxes,
some ovals, some with many different kinds of arrows and lines, with each notation
carrying a specific meaning. The notation used in the FRBR document diagrams
dates from the early days of E-R modeling. In this notation there are boxes for
entities and arrows for relationships. The arrows can have one head, meaning
that only one individual of the entity (e.g., one person) can be related, or they
can have two heads, meaning that the relationship can be “many” (figure 6.1).
75
©Karen Coyle CC-BY
76 / CHAPTER SIX
FIGURE 6.1
Many-to-many and one-to-many relationships in FRBR
WORK
The “Many-To-Many”
Relationship
is realized through
EXPRESSION
is embodied in
MANIFESTATION
The “One-To-Many”
Relationship
is exemplified by
ITEM
Looking at the E-R analysis technique in historical context alongside the work
of the Study Group that was developing FRBR helps explain what the diagrams
are capable of expressing. Entity-relation modeling and relational databases were
developed in the 1970s, and hit their peak in the 1980s. By 1990, the use of relational concepts was overlapping with a new computing paradigm: object-oriented
programming and database design. The FRBR group made use of an early version
of the E-R modeling concept and notation that was developed in the late 1970s.
By 1990, E-R modeling had added design features that allowed the expression of
more than just entities and relationships: in these new modeling notations it was
possible to indicate inheritance, precise cardinality, processes, and communication
paths. These later techniques would have made it possible to indicate whether
the FRBR entities were required or optional, something that is not included in
the FRBR entity-relation diagrams. Those diagrams show the same relationship
between persons and works as between expressions and works, yet we know
not every bibliographic description requires that a person be responsible for a
work, and the text of the FRBR Final Report states that every expression has a
mandatory relationship to a work. These are not distinguished in the diagrams.
By the time the FRBR Study Group provided its first draft in 1994, E-R modeling techniques had been replaced in technical design circles with the Unified
Modeling Language (UML), which was developed during the 1980s. UML is a
much more expressive language, with fourteen different diagram types, modeling
both structures and behaviors. It also is designed primarily for object-oriented
analysis, because entity-relation modeling had been superseded by object-oriented
design. Had UML been used by the FRBR Study Group the outcome of the
©Karen Coyle CC-BY
THE ENTITY-RELATION MODEL / 77
study might have been different, but that also would have required a different
skill set on the part of the Study Group members.
For a high-level view, a simple E-R model can still be useful. As its name implies,
E-R modeling views one’s information domain as entities or things, and defines
the relationships between those things. The E-R modeling technique provided
a structured approach for the FRBR Study Group, whose task was quite broadly
defined. Use of the technique was required by the Terms of Reference document
that gave the group its charge. In the Methodology section of the FRBR Final
Report, the group explains:
The methodology used in this study is based on an entity analysis technique that
is used in the development of conceptual models for relational database systems.
Although the study is not intended to serve directly as a basis for the design of
bibliographic databases, the technique was chosen as the basis for the methodology
because it provides a structured approach to the analysis of data requirements.
The FRBR Study Group makes clear that the resulting analysis is not a record
design, yet there is an acknowledgment that the FRBR Final Report answers
some questions that could be applied to bibliographic records:
The study makes no a priori assumptions about the bibliographic record itself,
either in terms of content or structure. It takes a user-focused approach to analyzing data requirements insofar as it endeavours to define in a systematic way
what it is that the user expects to find information about in a bibliographic record
and how that information is used. The study uses an entity analysis technique
that begins by isolating the entities that are the key objects of interest to users
of bibliographic records. (FRBR Final Report, 3)
However, the possibility that the analysis could be a precursor to database design
was also hinted at in the FRBR Final Report:
The entity-relationship analysis reflected in the model might also serve as a useful
conceptual framework for a re-examination of the structures used to store, display,
and communicate bibliographic data.
Barbara Tillett discusses this in her 1994 report on the work of the FRBR Study
Group: “We hope this exercise will provide the basics for development of future
structured bibliographic databases and future systems that facilitate creation,
maintenance, and use of such databases” (Tillett 1994).
©Karen Coyle CC-BY
78 / CHAPTER SIX
At the time, the E-R approach was new to at least some members of the FRBR
Study Group and was not part of most catalogers’ backgrounds. The FRBR
document itself refers to readings in this area that the Study Group members
found useful in understanding the entity-relation technology:
The entity-relationship analysis technique and the conventions for graphic presentation that are used in this study are based in large part on the methodology
developed by James Martin and outlined in his book Strategic Data-Planning
Methodologies (Prentice-Hall, 1982). Graeme Simsion’s Data Modeling Essentials
(Van Nostrand Reinhold, 1994), Richard Perkinson’s Data Analysis: the Key to
Data Base Design (QED Information Sciences, 1984), and Ramez Elmasri and
Shamkant Navanthe’s Fundamentals of Database Systems (Benjamin/Cummings,
1989) were also used in shaping the methodology for the study. All four books
are recommended to those who are interested in additional background and
more detail on entity-relationship analysis. (FRBR Final Report, 10)
Note the emphasis on database design. Also note the dates on the books cited—the
newest is now twenty years old. Relational database design, although still used in
business applications, is no longer cutting edge technology. Although modeling
of entities and relationships is still common, it has changed significantly from the
models in use in these books.
The FRBR entity-relation diagrams show a macro-level model that includes
only what are considered to be the primary relationships between entities. Other
relationships between bibliographic entities are defined in the text, such as work/
work relationships. These do not appear in the diagrams in the FRBR Final
Report, thus presenting an incomplete picture of the actual bibliographic model
described in the text.
ENTITIES, RELATIONS, AND DATABASE DESIGN
One of the reasons given for the development of an E-R model for bibliographic
data was the desire to create a bibliographic data model that was more in tune
with current technology. Because MARC was developed as a carrier for printed
bibliographic data, and preceded the automation of library catalogs, it wasn’t
designed with database technology in mind. That doesn’t mean that database
technology has not been employed in library systems; in fact, they would not
function as they do without the storage of data in such systems. Online systems
must make use of the efficiencies built into database management systems.
©Karen Coyle CC-BY
THE ENTITY-RELATION MODEL / 79
Restructuring MARC data for use in relational databases, as discussed in the
chapter on technology, is not an easy task. There are indeed some significant differences between bibliographic data and business data, and there is no question
that the main customer base for database technology is the business world, not
libraries. Therefore, the database technology that is on the market is optimized for
the needs of the majority (and richest) of customers. For example, bibliographic
data is primarily textual. Unlike much business data, bibliographic data has few
numerical amounts that need computation, and we know that computers are
more suited to work with numbers than with text. Also, there is not an even or
predictable amount of repetition in bibliographic data; there are some authors or
subjects that have high redundancy in a file, but there are even more that exist in
a single exemplar. Relational databases are at their most efficient when the same
data repeats frequently in the database, but provide less of an advantage for data
with a high level of uniqueness.
In spite of the fact that bibliographic data isn’t what database management
system developers had in mind when developing the technology, from the very
first every library system has made use of some of the features of a database
management system in order to function. It is therefore a misunderstanding to
assume that because library data is not easily normalized into ideal relational
database forms, library systems do not make use of relational database technology.
They do, although the result does not look like the idealized design in database
design textbooks.
As part of the development of the FRBR-informed cataloging rules, Resource
Description and Access (RDA), Tom Delsey developed the RDA Implementation
Database Scenarios, which depict “database structures conventionally used in
library applications” (Delsey 2009). Scenario 3 shows library systems working
with MARC-based data using a “flat file” approach, which would be similar to
data stored in a spreadsheet. In fact, no flat file–based system could produce the
kind of retrieval that library systems provide, and most systems today are at least
as sophisticated as that document’s Scenario 1, which appears to be the preferred
model for the management of RDA data. The virtues of the MARC record, with
its variability of field and subfield combinations and the unlimited repeatability
of most fields, make it unsuited to a flat-file treatment. It would not be possible
to provide search or browse on field types, like titles or subjects, without making
use of entities and relations. In fact, the database I worked on in the early 1980s
definitely used a relational design. A mock-up of its very high-level design is shown
in figure 6.2. There were also many other indexes for corporate authors, titles,
dates and languages, which I don’t show here for reasons of space. An actual
©Karen Coyle CC-BY
80 / CHAPTER SIX
database design is a mass of boxes and arrows that often cannot be reproduced
on a single sheet of paper.
FIGURE 6.2
Inside A library system database, circa 1984
SUBJECT INDEX
SUlink1
PERSONAL AUTHOR INDEX
Mentally ill — Fiction
PAlink1
Melville, Herman
PAlink2
Olmstead, Kathleen
BIBLIOGRAPHIC DATA
100
[PAlink1]
245
Moby Dick
300
356 p., illus
650
[SUlink1]
700
[PAlink2]
There are a couple of points that should be taken away from this. One is that
although a data structure that has clear entities and relations defined may be
somewhat easy to extend into a database design, many times a database design is
derived from data that was initially developed for some other purpose. Although
the bibliographic data that is stored in the MARC record still adheres to a structure
that originally supported the card catalog, an E-R analysis can be done that results
in a database that supports search and display of the data. This database design
is primarily based on practical considerations: enabling retrieval of headings and
combinations of headings with fast response time, even within large databases.
Although the database model for bibliographic data differs considerably from,
for example, that of banking, library systems run on the same underlying technology, making use of the features that a database management system provides.
This means that the FRBR E-R model is not the first practical use of E-R
modeling for bibliographic data. The opportunity that FRBR afforded was a
rethinking of bibliographic data model with entities and relations in mind, which
did not adhere to the model of description and headings that has been the form
of bibliographic data for centuries. The goals stated in the document— facilitating
sharing and decreasing the costs of cataloging on an international scale—may
have motivated the FRBR Study Group to develop the entities and relationships
in FRBR, although the connection between the goals and the E-R model are
not presented clearly in the FRBR Final Report.
©Karen Coyle CC-BY
S E V EN
W H AT I S
MODELED IN
FRBR?
A
n E-R analysis serves to resolve category boundaries and assign attributes to categories of things or functions. That said, for any given data,
there can be any number of E-R models developed, depending on the
functionality desired, the requirements of your data management system, and
the workflow you need to support. The same is true for bibliographic data. The
top-level model developed by the publishing industry has three primary “things”:
people, stuff, and deals (figure 7.1).
This represents a bibliographic model that primarily supports commercial
functions around intellectual resources. The library model developed as FRBR
could be described as “people, stuff, and subject access.” Each model reflects the
needs and views of its community.
The simpler your goals, the simpler your data model can be. However, the
FRBR Study Group had a rather complex set of goals. One goal had to do with
simplifying the bibliographic record for international sharing, with the purpose
of cost savings. Another goal required the Study Group to make a connection
81
©Karen Coyle CC-BY
82 / CHAPTER SEVEN
FIGURE 7.1
<indecs> commercial view of intellectual products
make
PEOPLE
STUFF
used by
do
about
DEALS
between user needs and bibliographic data elements. This was related to the first
goal, because the data elements most needed by users would also be the ones
that truly should be in the data model. There also seem to have been goals that
group members brought to the effort, such as codifying bibliographic relationships between described resources. The use of E-R modeling was itself a goal,
which was possibly included because some members assumed that a future bibliographic record would be stored in relational databases, or that sharing would
be easier if the data were packaged as separate entities. Because the FRBR Final
Report doesn’t address technical issues or a record format, those goals are not
clarified in the report.
FRBR defines three groups of entities. The groups are not named; they are
called only Groups 1, 2, and 3. The groups themselves are included in some of
the E-R diagrams as boxes around the entities of the group, but are ignored in
further modeling. This means that there are no functions or qualities that belong
to the groups themselves. Each entity is treated as separate. There are no group/
entity relationships that would create a type of class/sub-class structure. There is
also no entity or identified class that represents a whole bibliographic description.
This brings up the question of whether the groupings of entities in FRBR are
meaningful at all. Gordon Dunsire, who has created the FRBR representations
in RDF for the IFLA FRBR Review Group, appears to have been instructed that
the groups are not to be used as classes in the RDF sense.
FRBR Group 3 is not represented in RDF as a class, following clarification from
the FRBR Review Group: the Groups are used to simplify the entity-relationship diagrams, and are not intended to be super-classes. Instead, 10 separate
properties are represented in RDF, all with domain Work and each with one of
©Karen Coyle CC-BY
WHAT IS MODELED IN FRBR? / 83
the Group 1, 2, or 3 entity classes as range, corresponding to Figure 3.3 in the
FRBR report. (Dunsire 2012, 736)
Each of the FRBR groups has a different conceptual structure. Group 1, which
could be thought of as representing a bibliographic description, consists of four
mutually dependent entities that are modeled as a chain from work to expression to manifestation to item. Although the diagram does not specify whether
the entities and relationships are mandatory, it is clear to most readers that all
four are needed for a complete bibliographic description, and that having, for
example, an expression entity with no work or manifestation entity would not be
meaningful. It isn’t clear, however, whether the model intends to make all four
mandatory as part of a description.
Group 2 consists of person and corporate body, and these have agent- or actortype relationships with entities in Group 1. The Group 2 entities, unlike the
Group 1 entities, have no relationships that link them to each other analogous to
the intra-group relationships of Group 1. They also do not share any attributes.
They could have been modeled as members of a class because logically they do
share some attributes, like the relationships linking them to the Group 1 entities.
Both of these entities could be an author, a publisher, a performer, etc., and
therefore those attributes could be assigned to a class that includes all Group 2
entities, but they were not.
Group 3 includes four entities that can have a “subject” relationship to the
work entity in Group 1: concept, object, event, and place. These, too, have no
links between them and are not members of a mutual class. Although these four
entities are a group called “Subject” in the FRBR text, in fact all entities from
Groups 1 and 2 can also have a subject relationship with a work entity. This means
that logically all FRBR entities could be sub-classed to a subject class. Group 3
appears to round out the entities needed for subject assignment, but isn’t itself a
complete list of subject types even though it is referred to as the subject group.
FRBR AS A CONCEPTUAL MODEL
FRBR is not a data model. FRBR is not a metadata scheme. FRBR is not a
system design structure. It is a conceptual model of the bibliographic universe.
(Tillett 2005)
The name “Functional Requirements for Bibliographic Records” evokes a much
more concrete outcome than was actually presented in the FRBR Final Report.
©Karen Coyle CC-BY
84 / CHAPTER SEVEN
Functional requirements generally speak to actions, workflows, and methods.
A bibliographic record, at least as most of us think of it, is a defined set of data
elements that identify and describe a resource. The entity-relation model is also
intended to provide a workable model that could eventually be instantiated in
some type of computer application. However, the result of the FRBR Study
Group’s work has often been described as something much less concrete than
the name might imply, that is, a conceptual model.
The FRBR Final Report’s section 2.3 on methodology gives the reasoning
behind the use of entity-relation modeling technique:
The methodology used in this study is based on an entity analysis technique
that is used in the development of conceptual models for relational database
systems. Although the study is not intended to serve directly as a basis for the
design of bibliographic databases, the technique was chosen as the basis for the
methodology because it provides a structured approach to the analysis of data
requirements that facilitates the processes of definition and delineation that were
set out in the terms of reference for the study. (FRBR Final Report, 9)
Entity-relation modeling is a multistep technique that begins with a high-level
conceptual analysis of the data universe that is being considered. To quote once
again from the FRBR Final Report:
The first step in the entity analysis technique is to isolate the key objects that
are of interest to users of information in a particular domain. These objects of
interest or entities are defined at as high a level as possible. That is to say that
the analysis first focuses attention not on individual data but on the “things” the
data describe. Each of the entities defined for the model, therefore, serves as the
focal point for a cluster of data. An entity diagram for a personnel information
system, for example, would likely identify “employee” as one entity that would
be of interest to the users of such a system. (FRBR Final Report, 9)
This is a very good description of conceptual modeling. So it is either puzzling or
disturbing that most readings of FRBR do not recognize this difference between
a conceptual model and either a record format or a logical model. In part this
is because it is easy to view the diagrams in the document as statements of data
structure rather than high-level concepts about bibliographic data. This may also
be because most members of our profession are not familiar with the stages of
modeling that are used in formal database design.
©Karen Coyle CC-BY
WHAT IS MODELED IN FRBR? / 85
One of the common assumptions about FRBR is that the entities listed there
should be directly translated into records in any bibliographic data design that
intends to implement FRBR. For example, there is much criticism of BIBFRAME
for presenting a two-entity bibliographic model instead of using the four entities
of FRBR. This reflects the mistaken idea that each Group 1 entity must be a
record in whatever future bibliographic formats are developed. As entities in a
conceptual model oriented around database design, there is absolutely no direct
transfer from conceptual entities to records. How best to create a record format
that carries the concepts is something that would be determined after a further
and more detailed technical analysis. In fact, the development of a record format
might not logically be a direct descendent of the E-R model, because the E-R
modeling technique has a bias toward the structure of relational database management systems, not records. In addition, should a conceptual design like FRBR
be used to inform the next steps toward a database design, there is no guarantee
that the final design will retain the high-level structure of the conceptual model.
Few assumptions can be made about the potential technical implementation
based on a conceptual design; only further analysis, with a specific technology
as its target, can reveal that.
One also cannot make assumptions about record design based on a database
design model. For many databases there is no single record that represents all
of the stored data. Databases are often a combination of data from numerous
departments and processes, and they can receive and output many different data
combinations as needed. The database does not define the record format although
they must share the definition of the atomic elements that both carry.
The FRBR Final Report has a section recommending areas for further study.
In that section the FRBR Study Group states that the report “is intended to
provide a base for common understanding and further dialogue, but it does not
presume to be the last word on the issues it addresses.” Areas for further study
include expanding the analysis to authority data, which is being taken up by the
ongoing working group within the IFLA Cataloguing Section. The report also
suggests performing studies to verify the validity of the attributes listed for the
FRBR entities. And finally, it addresses the potential of the entity-relation model
to inform a new record format:
The entity-relationship analysis reflected in the model might also serve as a
useful conceptual framework for a re-examination of the structures used to store,
display, and communicate bibliographic data. Further study could be done on
the practical implications of restructuring MARC record formats to reflect more
©Karen Coyle CC-BY
86 / CHAPTER SEVEN
directly the hierarchical and reciprocal relationships outlined in the model. An
examination of that kind might offer a new approach to the so-called “multiple
versions” issue. The model could also be expanded in depth to create a fully
developed data model that would serve as the basis for the design of an experimental database to assess the efficiency and effectiveness of a database structure
patterned on the model. (FRBR Final Report, 6)
This recommendation has not had follow-up, and most likely will not, at least
not in the sense described here. By the time that the FRBR Final Report was
issued in 1998, relational models were on the wane. By 2008, even the successor
technology, the object-oriented model, was being supplanted by the data design
concepts of the Semantic Web and linked data. Had the library world embraced
a relational data design by the end of the 1980s, library data and library systems
might have been in line with common information technology development. As
it is, the time for a relational design has passed.
GROUP 1 WORK, EXPRESSION, MANIFESTATION, ITEM
Group 1 of FRBR comprises the main entities of bibliographic description: work,
expression, manifestation, and item (WEMI). There is a set of relationships
between these entities, and they are described in a linear pattern. The FRBR
Final Report presents the entities from work to item, moving from the most
abstract to the most concrete. One could also take the view of the cataloger’s
workflow, which begins with the item in hand, and moves through manifestation and expression to work. However, FRBR does not describe the process of
creation, but rather a fully realized resource. There is no temporal order implied
between the entities of Group 1.
It is the relationships between these entities that complicates Group 1 and
also that leads to different interpretations of what possibilities exist to make use
of the different combinations of entities. It also creates some complication in
the relationships with other entities because each Group 1 entity has separate
bibliographic relationships with other FRBR entities. This means that there are
not only relationships between work, expression, etc., but there are so-called
“bibliographic relationships” between works, between expressions, between
works and expressions, and so on. The resulting picture is of a very complex
web of relationships.
©Karen Coyle CC-BY
WHAT IS MODELED IN FRBR? / 87
The FRBR Work
As we saw in a previous section, numerous definitions of “bibliographic work”
have been developed in the library field over time. FRBR defines the work entity
as “a distinct intellectual or artistic creation.” In reference to FRBR work, Barbara
Tillett, in her 2003 Library of Congress Pamphlet What Is FRBR? defines work
as “the conceptual content that underlies all of the linguistic versions, the story
being told in the book, the ideas in a person’s head for the book.” The work
as idea is however not entirely borne out by the Final Report text. The FRBR
Final Report declares changes in form (e.g., from a book to a movie version of
the same story) to be different works, which hints at a more precise definition of
work than being a story or idea. This illustrates how difficult it will be to form a
widely accepted definition of work. In fact, the FRBR Final Report acknowledges
that work may be defined different in different communities.
Because the notion of a work is abstract, it is difficult to define precise boundaries
for the entity. The concept of what constitutes a work and where the line of demarcation lies between one work and another may in fact be viewed differently from
one culture to another. Consequently the bibliographic conventions established by
various cultures or national groups may differ in terms of the criteria they use for
determining the boundaries between one work and another. (FRBR Final Report)
The FRBR Final Report also states that “We recognize the work through individual realizations or expressions of the work, but the work itself exists only in
the commonality of content between and among the various expressions of the
work.” This is considerably different from the definitions given in the previous
paragraph. In this definition, the work is a set of all of its expressions. This would
make the work a dynamic entity that changes depending on the presence of
particular expressions in the bibliographic universe, similar to Patrick Wilson’s
bibliographic families that grow as new family members are born. Thus the work
is presented both as a fixed a priori abstraction (in the mind of the creator) and
as a dynamic entity that is the sum of its expressions.
The work is consistently referred to as an abstraction in library literature about
FRBR, but the work as defined in FRBR has the characteristics of something
quite concrete: it has a creator, a genre, and subject designation, which is actually
quite a bit of “realness” for something that is “an idea in a person’s head.” It also
has a work title, which could be considered especially odd for something that
has not be expressed or manifested yet. However, the work title (along with the
©Karen Coyle CC-BY
88 / CHAPTER SEVEN
creator) serves as the bibliographic identifier for the work, so it has a necessary
bibliographic purpose. This is an important point: the definition of the entities
in FRBR is not the expression of a philosophical or theoretical declaration of a
bibliographic ideal; it is fully grounded in library bibliographic practices, and the
model is intended to support those practices.
The FRBR work has many potential relationships with other entities. It has
a primary relationship with one or more expressions. (Primary relationships are
defined in the FRBR Final Report as the ones visible in the entity-relation diagrams.) It also has primary relationships with persons and corporate bodies. It is
the only entity that has a “has subject” relationship with other entities.
The FRBR Expression
The FRBR expression is “the intellectual or artistic realization of a work in the
form of alpha-numeric, musical, or choreographic notation, sound, image, object,
movement, etc., or any combination of such forms” (FRBR Final Report, 19).
The expression is also an abstract entity in that there is no physical realization
of the expressed work until it is manifested in a physical format. It has a form of
expression (e.g., “text” or “music”) and its attributes include a date, a language
(for textual works), a medium of performance (for music or other performed
works), and scale and project (for cartographic works). The expression excludes,
however, “physical form . . . that [is] not integral to the intellectual or artistic realization of the work as such” (FRBR Final Report, 19). Therefore, the expression
is the artistic realization of work in every aspect except the physical form. It does
include “the specific words, sentences, paragraphs . . . or the particular sounds,
phrasing, etc. resulting from the realization of a musical work.” The expression
has all the qualities of a creation, but stops short of any physical attributes. So
it is a text, but without a “typeface and page layout.” It is music without any
readable notes or carrier of sound.
Here we immediately face one of the key issues of the FRBR Group 1 entities:
what is the nature of the Group 1 entities in relation to what is essentially a single
thing that has been described in four parts? The expression is a realization of a
work, and as a realization of a work it not only is not separate from it, it actually is
the intellectual content of the work in some form. Although in the entity-relation
diagrams and in the lists of attributes the work and the expression are treated as
separate, conceptually the expression cannot be separated from the work because
the work is the intellectual content of the expression.
©Karen Coyle CC-BY
WHAT IS MODELED IN FRBR? / 89
The expression is generally considered the most conceptually elusive of the
Group 1 entities. Although the work is abstract, it has creative content, creators,
and subjects. The manifestation has a physical description and most of us have considerable experience with manifestations in our lives. The expression is described
as a text that has no particular physical form, something that seems impossible
because just calling it a text must mean that it exists in some form or another.
The expression is also potentially the result of an editorial process or some other
third-party contribution, because the FRBR Final Report speaks of expressions
with glosses, illustrations, and augmentations. For anyone who envisions WEMI
as a directional flow from work to item, or vice-versa, expression is a sticking
point because it doesn’t have a single direction. The expression is the entity of
translations of texts and performance of music, but also of annotated editions
and selected works.
It is worth noting here that there is no concept of the creator’s original
expression in FRBR. All expressions have equal weight. This is potentially an
area where Patrick Wilson’s second kind of power would come into play, and
some qualitative assessment of expressions, along with a family genealogy, could
be offered to users seeking the best content for their need.
The FRBR Manifestation
The FRBR manifestation is “the physical embodiment of an expression of a work”
(FRBR Final Report, 20). Like the expression, the manifestation has content that
has carried over from the preceding entities. A manifestation may be a unique,
stand-alone object, as in an author’s manuscript, or it can be a set of mass-produced
things made publicly available, or it can be anything in between. The physicality
of the manifestation is a primary characteristic:
When the production process involves changes in physical form the resulting product is considered a new manifestation. Changes in physical form include changes
affecting display characteristics (e.g., a change in typeface, size of font, page layout,
etc.), changes in physical medium (e.g., a change from paper to microfilm as the
medium of conveyance), and changes in the container (e.g., a change from cassette
to cartridge as the container for a tape). (FRBR Final Report, 22)
This brings us squarely into the “multiple versions” territory, one of the compelling cataloging issues of the 1990s: does a change of physicality require a new
©Karen Coyle CC-BY
90 / CHAPTER SEVEN
catalog record? The answer in the FRBR Final Report is clearly that a change in
physical form is a different manifestation. Therefore a hardback book and a trade
paperback of the same book would be different manifestations, and each e-book
format would also be a different manifestation. This would be in keeping with
the distinctions managed by publishers of these materials, and in the assignment
of a specific ISBN to each physical format. However, considering that resources
in digital formats, such as electronic books, can be converted to different digital
formats at the time of access, it could be said that the physical stability of the
manifestation is becoming a thing of the past.
Note that manifestation is not separate from the expression of the work as
described here; it embodies it. In fact, the FRBR Final Report says that the manifestation embodies an expression of a work, and therefore is defined as the whole
Group 1 minus the item. That it embodies makes it sound like the manifestation
is a kind of container for the expression and the work. That it manifests instead
sounds more like it manufactures or creates. In addition, the manifestation contains certain elements provided by the publisher, like a title page, often an index,
and in some cases some prefatory material, which then brings into question if
or when the publisher’s additions constitute the creation of a new expression.
In some instances, any one particular expression is only one of the creations
that is manifested in a publication. Such publications are referred to in FRBR as
“aggregates.” Although addressed in the FRBR Final Report, aggregates pose
some difficult problems and had to be addressed by a special sub-group of the
FRBR Study Group. At this point it should be obvious to the reader that the
four boxes and few arrows of the diagrams in the FRBR Final Report are a gross
simplification of the bibliographic domain that FRBR attempts to capture.
The FRBR Item
Item is the most neglected of the Group 1 entities in most discussions, possibly
because it is also the most clear. “The entity defined as item is a concrete entity”
(FRBR Final Report, 24). An item exemplifies a single manifestation. However,
even here the meaning has some caveats. An item may consist of more than one
physical object, such as a multivolume monograph, or a film or sound product
issued on multiple disks. Each item exemplifies a single manifestation. In certain
cases, such as with copies of special interest, there may be some variation between
a single item and other items of the same manifestation, such as books that are
signed by the author, or items that are damaged.
©Karen Coyle CC-BY
WHAT IS MODELED IN FRBR? / 91
GROUP 2
PERSONS, CORPORATE BODIES, FAMILIES
In the FRBR Final Report, Group 2 has two entities, person and corporate
body. Subsequent work on the Functional Requirements for Authority Records
(FRAD) has added a third entity, family, which is now assumed to be included
in FRBR. The FRBR document does not add any new features to persons and
corporate bodies compared to current authority records, although FRAD does
expand the attribute list to include others, such as gender and places of birth,
death, and residence for persons, and language, history, and field of activity for
corporate bodies.
As with the other groups, there is no whole that would unify Group 2 as
a meaningful unit, even though the entities are treated as a single unit in the
text. Without a more general class or entity to which these could belong, some
characteristics, like roles, will logically need to be defined separately for each
entity, with a significant amount of redundancy. For example, both persons and
corporates bodies can be publishers, collectors, dedicatees, defendants, and so
on. The lack of a super-class that unifies Group 2 means that it will be necessary
to decide which roles are appropriate for each Group 2 member. As we will see,
those implementing FRBR as a conceptual model usually develop a super-class
for the entities in this group, which also enables working with data where the
identity of the agent (especially person or corporate body) has not been clearly
determined.
GROUP 3
SUBJECTS
The nature of the third FRBR group is something of a miscellany: concept,
object, event, and place. It isn’t made clear in the text how FRBR Study Group
members arrived at these entities. They do not match the elements of the Library
of Congress Subject Headings, which are topic, form, chronology, and place.
While Group 3 is known as the subject group, it is not the full range of subjects
in FRBR. All of the FRBR entities can be in a subject relationship to the work.
The diagram also depicts the “subject” relationships between work and the
entities in the first and second groups. The diagram indicates that a work may
have as its subject one or more than one work, expression, manifestation, item,
person, and/or corporate body. (FRBR Final Report, 17)
©Karen Coyle CC-BY
92 / CHAPTER SEVEN
Robert Maxwell (2008) notes in his book on FRBR that the limitation of subject
relationships to the work entity ignores some cases in which expressions or even
manifestations can have a need for topical access on their own. He suggests that
“it might be preferable in FRBR simply to define Group 3 entities as ‘entities that
serve as subjects of Group 1 entities’” (Maxwell, 15). Actually, that definition
would also have to be applied to Groups 1 and 2, because all entities can be the
object of a “has subject” relationship.
We don’t learn any more about the Group 3 because the only attribute provided for each of the entities is “term.” The document expressly limits the use
of the Group 3 entities to subject access, although in the introduction there is
an acknowledgment that an entity like “place” might have use within the bibliographic description, because it would also be logical for an object to have a
place, and for that place to have a name and a geographical location with longitude and latitude.
We also learn little about the use of these entities, because other than defining
these entities and showing an E-R diagram with the subject relationship, subjects
are not mentioned in the document. Quite astonishingly, they are not listed in
the discussion of the user tasks. The find task lists:
the attribute or relationship is typically used as a primary search term for locating
the entity (e.g., the title of a manifestation, the relationship between a work and
the person or corporate body principally responsible for the work). (FRBR Final
Report, 84)
We should remember that subjects were not initially included in the Terms of
Reference, but were added to the list of entities when the document was reviewed
at the IFLA meeting in 1992. It is evident that there is a strong separation between
descriptive cataloging and subject cataloging in library practice, and this has an
effect on the FRBR Study Group’s attention to the subject aspect of user needs.
This points out a chasm between library cataloging and user service that FRBR
does not address.
THE BIBLIOGRAPHIC RELATIONSHIPS
Nearly all attention on FRBR today focuses on the ten (or eleven, if you count
family) entities. Little attention is given to the bibliographic relationships that
FRBR defines. Although they echo bibliographic relationships in current cataloging, in FRBR each relationship is designated as being between specific FRBR
©Karen Coyle CC-BY
WHAT IS MODELED IN FRBR? / 93
Group 1 entities. These bibliographic relationships are distinguished from the
primary relationships that exist between the FRBR entities, and which are shown
in the diagrams in the FRBR Final Report. The bibliographic relationships,
when added to the basic FRBR entity-relation diagram, create a very complex
web of connections.
Both Barbara Tillett (1988) and Richard Smiraglia (2001) conducted quantitative studies of the relationships that exist in library catalogs. Tillett did her study
on a large portion of the Library of Congress MARC file. Smiraglia researched
the catalogs of a small number of research institutions, including two specialized collections, one of theology and one of music, and focused on works and
work families (the latter included many relationships that FRBR would define
as between expressions). Both found a high number of related resources in the
catalogs they studied, although exact numbers are difficult to assign because
making bibliographic relationships manifest is optional in library cataloging.
Relationships are generally provided only in notes, and only if the cataloger deems
that the note is needed for clarity.
It isn’t clear if the presence of bibliographic relationships in FRBR implies that
they are to be coded as actual links between descriptions, or if they are simply a
conceptual rendering of the idea of bibliographic relationships. If one decides that
the relationships should be treated as links between entities, then the optionality
of explicit relationships described in the report becomes problematic because not
all related entities would be linked. In section 7 of the FRBR Final Report, where
the basic level record is defined, each of the notes that defines a bibliographic
relationship has a footnoted explanation to the effect that such notes are made
only if the relationship cannot be inferred from other information in the record.
However, if the relationship, as implied in the use of an information technology
model for the entities and relationship, is intended to be actionable in a database or other information system, there is no means to “infer” relationships; all
data to be acted on must be explicitly coded. Thus, the assumption that FRBR
represents a machine-actionable model is clearly contradicted in this area of the
FRBR Final Report, leaving us to wonder exactly how we should interpret the
use of the entity-relation model vis-a-vis the catalog record data.
Summary
Group 1, as defined in the FRBR Final Report, exhibits some distinct differences
between how it is defined in the text and how it is depicted in the entity-relation
diagram. The entity-relation diagram shows four separate entities, and these entities are given each a separate set of attributes. This is consistent with a relational
©Karen Coyle CC-BY
94 / CHAPTER SEVEN
database model that creates separate entities based on an analysis of redundancy
of data elements. This does not mean that those entities are separate in “real life.”
We can use the example of a car model and color options. Obviously there is no
car being offered that has no color, but color options can be stored in a separate
table from car models in a database so that models and colors can be combined
as needed in database output. This is similar to the separation of attributes into
the WEMI entities in FRBR.
At the same time, the text of FRBR describes the entities as realizing or
embodying or exemplifying other entities, which implies less separation than the
entity-relation diagrams describe. Conceptually, an item is a concrete example of a
manifestation that embodies the expression of a work. When you hold the item in
your hand, you are holding something that has within it an entire Group 1. This
statement results in a very different diagram from the entity-relation diagrams
in the FRBR Final Report.
FIGURE 7.2
E-R diagrams versus definitions in the text
FRBR E-R MODEL
FRBR AS DEFINED IN TEXT
ITEM
WORK
is realized through
MANIFESTATION
embodies
MANIFESTATION
is exemplified by
EXPRESSION
expresses
is embodied in
WORK
realizes
EXPRESSION
ITEM
Both of these pictures of WEMI can be valid because they represent different
views of the same thing. On the left of figure 7.2, we have a diagram of how
WEMI will be expressed as stored metadata, in particular metadata that has been
subjected to an entity-relation analysis. On the right, we have a diagram of a
conceptual view of those same entities, without the need to separate the entities
for the purposes of data efficiency. In fact, the diagram on the right, if it were
©Karen Coyle CC-BY
WHAT IS MODELED IN FRBR? / 95
realized as a data format, would entail a great deal of redundancy of data. It is,
by the way, very close to the data format that we use today. Which brings us back
to FRBR as an entity-relation analysis of bibliographic data as it was created in
libraries at the end of the twentieth century. If we add the two main authority
files to the diagram on the right, we essentially have the equivalent of FRBR’s
entities prior to an entity-relation data analysis.
Looking at FRBR from this same point of view, it is clear that Group 1 represents a single thing, but that thing can be separated into parts for the convenience
of managing nonredundant attributes. FRBR also describes the elements of a full
bibliographic description. It is less clear, though, how those two views interact
in practice, whether in the act of cataloging an item in hand, in sharing cataloging data, or in serving the library’s users. Because FRBR was instantiated in the
twenty-first-century cataloging rules, Resource Description and Access (RDA),
the first of those questions may have been answered. The other questions, at the
time of this writing, are still open.
©Karen Coyle CC-BY
©Karen Coyle CC-BY
E IG HT
DOES FRBR
MEET FRBR’S
OBJECTIVES?
Many people want to evaluate a conceptual model like FRBR using “true or false”
criteria. While one can say that a model is “true” to the extent that it explains
accurately and “false” to the extent that it does not, this is not a very helpful
way to look at these models. A more useful way to evaluate these models is to
ask whether they are successful at fulfilling their purposes.
—Alison Carlyle, 2006, 266
T
he FRBR model has resulted in unprecedented change in our thinking
about catalogs and cataloging. Since the mid-nineteenth century, cataloging has focused on a unitary description of a physical item and a primary
trinity of access points: title, author, and subject. This model has been disrupted
by new technologies of communication, from the recording of sound and moving
pictures to the digitization of nearly everything. Added to that disruption is the
ever-increasing rate of production of creative resources of all types.
97
©Karen Coyle CC-BY
98 / CHAPTER EIGHT
The previously stable view of the role of library catalogs began to be challenged
in the latter half of the twentieth century as libraries struggled with these changes
in their holdings. The key blow to the library’s stability, however, was wrought by
the Internet, the global network that allows instantaneous worldwide publication
that defies the barriers of time and place. The slow conversation of books, which
allows time for research, synthesis, creation, and production, is giving way to fast,
short, mashed-up, and transitory personal and cultural expression.
Fortunately, some of the causes of this change are also potential factors in
managing the change. Without computer technology libraries would not have
been able to quickly add shared bibliographic records to their catalogs and thus
to keep up—to the extent that they have—with the rate of resource production.
The FRBR model developed by the FRBR Study Group is that group’s response
to some of the late twentieth-century challenges. One can see in the conceptual
model a strategy of “divide and conquer,” a completely different approach from
the “publication as unit of interest” of the cataloging concepts that guided the
development of the AACR. In his essay in the collection The Future of the Descriptive Cataloging Rules, “AACR3? Not!” Michael Gorman, editor of AACR2,
disputes the cataloging compromises proposed as a solution to the problem of
multiple physical formats for the same content, saying that “Descriptions are of
physical objects. . . . It is literally impossible to have a single description of two
or more different physical objects” (Gorman 1998, 27).
FRBR is a direct challenge to the emphasis on a single unit defined by its
physical description in descriptive cataloging rules. Although it doesn’t ignore
physical description, it does relegate it to a secondary role behind the placement
of the content in a larger bibliographic context. FRBR introduces abstractions
for works and expressions that had only appeared in earlier catalogs as collocation
mechanisms, such as uniform titles.
As Carlyle says, though, the measure of success of a model is not its truth or
falsity, but whether it achieves its goals. The objectives that were put forth by
the Terms of Reference that guided the work of the FRBR Study Group give us
one possible way to measure FRBR’s success. This section looks at FRBR and
its objectives, and asks:
``
``
``
``
``
``
Does FRBR provide functional requirements?
Does FRBR produce a national level bibliographic record?
Does FRBR serve user needs?
Does FRBR promote cataloging efficiency?
Does FRBR aid data sharing?
Is FRBR format neutral?
©Karen Coyle CC-BY
DOES FRBR MEET FRBR’S OBJECTIVES? / 99
DOES FRBR PROVIDE FUNCTIONAL REQUIREMENTS?
To begin with, we must note that our rules of descriptive cataloging contain no
statement of the function which they are designed to serve. (Seymour Lubetzky,
1946, “Analysis of Current Descriptive Cataloging” 1946)
The study has two primary objectives. The first is to provide a clearly defined,
structured framework for relating the data that are recorded in bibliographic
records to the needs of the users of those records. The second objective is to
recommend a basic level of functionality for records created by national bibliographic agencies. (FRBR Final Report)
Seymour Lubetzky had two primary complaints about the cataloging rules and
their outcome: first, that the rules stated what decisions catalogers should make,
but not why the elements of the catalog were necessary; second, that the information in the catalog entry was not presented to the user in the order of importance,
thus requiring the catalog user to wade through information of lesser importance
in order to find those elements that were relevant. Both of these concerns speak
to the role of the catalog in serving the needs of the user. Although Lubetzky
did not articulate what those important elements were, it is easy to imagine that
a user might find more value in an author added entry than the height of the
book in centimeters.
One of the primary goals of FRBR was to rectify the lack of a functional justification by clearly aligning the data of the bibliographic record with the functions
of the record and how it is used. This is the “functional requirements” of FRBR,
and one cannot ignore that a stated purpose of the FRBR Study Group was to
develop those functional requirements. It is also impossible to ignore the fact that
functionality from the user perspective plays only a minor role in the final report.
Much is made of the user tasks find, identify, select, and obtain that are introduced in the FRBR Final Report. They are described there as “generic tasks” and
are not made more specific. The description of them does not obviously motivate
the selection of entities nor their attributes. In fact, the analysis of the user tasks
is presented in the Final Report only after the entities and their attributes have
been defined in some detail. It is significant that the document’s structure is
organized around the groups of entities, not the user tasks. The user tasks are
covered in chapter 6 of the document, and are presented almost entirely as a
comparison of the user tasks and the attributes that have been defined for the
entities that are the main focus of the document. Attributes are rated as having
a high level, medium level, or low level of importance for each of the broad user
©Karen Coyle CC-BY
100 / CHAPTER EIGHT
tasks, as shown in figure 8.1. Some of the attributes are only of low importance,
which brings into question why they are even included as necessary attributes.
For example, the Work/Work relationships summarization, adaptation, transformation, and imitation are listed as low value for identify and select. Had the
focus of the report truly been user needs, it is doubtful that those attributes
would have been included.
FIGURE 8.1
A segment of a table of user tasks and FRBR attributes
Work
Expression
Manifestation






Statement of responsibility
  





ATTRIBUTES OF MANIFESTATION
Edition/issue designation
Place of publication/distribution
Publisher/distributor
Date of publication/distribution




  
 

Fabricator/manufacturer (note 1)


Series of statement


Form of carrier


Extent of the carrier (note 2)

Physical medium (note 3)

Capture mode

Dimensions of the carrier (note 4)

Manifestation identifier



©Karen Coyle CC-BY
Item
Manifestation
Title of the manifestation
Item
Expression
IDENTIFY
Work
FIND
DOES FRBR MEET FRBR’S OBJECTIVES? / 101
The user tasks cover only the so-called “primary” entities, those of Group 1.
There is no mention of how a user knows that she has found the correct entry for
an author or a subject. In fact, both authors and subjects get short shrift throughout the FRBR Final Report, and it is presumed that they will be described more
fully in the companion standards, Functional Requirements for Authority Data
(FRAD) and Functional Requirements for Subject Authority Data (FRSAD).
FIGURE 8.1
A segment of a table of user tasks and FRBR attributes
ATTRIBUTES OF MANIFESTATION
Title of the manifestation
  

Statement of responsibility

Edition/issue designation
Place of publication/distribution
Publisher/distributor
Date of publication/distribution






  

 

 


Fabricator/manufacturer (note 1)

Series of statement


Form of carrier


Extent of the carrier (note 2)
 
Physical medium (note 3)

Capture mode

Dimensions of the carrier (note 4)

Manifestation identifier


©Karen Coyle CC-BY
Item
Manifestation
Expression
Work
OBTAIN
Item
Manifestation
Expression
Work
SELECT
102 / CHAPTER EIGHT
Let’s look at just a few aspects of the user task analysis. For the task find, the
attributes of high value to find a work are:
``
``
``
``
``
``
title of the work
dependent component
independent component
persons/corporate bodies responsible for work
entities treated as subject of work
title of the manifestation
Attributes of high value to find a manifestation are:
``
``
title of the manifestation
manifestation identifier
Because no detailed explanation was given for these specific selections, one can
only surmise what the FRBR Study Group was thinking. The title of the work, the
“responsible parties,” and the subjects are unsurprising, and mirror the ages-old
goals of the catalog first expounded by Cutter. Less obvious are the dependent
and independent components. These are essentially whole/part relationships,
in which a dependent component would be a chapter in a book, whereas an
independent component would be a monograph in a monographic series, or an
article in a journal. These are real relationships, but because very few dependent
parts are cataloged, the importance of these for the find task may be limited.
As for the independent components, these are usually sought on their own, as
in the case of journal articles. A direct relationship between the larger unit and
the part is obviously a good idea, as primary elements for finding materials these
aren’t convincing.
It is also interesting that the manifestation title is an important element for
finding the work, but the work title is not equally important for finding the manifestation. There is undoubtedly some logic behind that, but it is not explained.
For the manifestation, the “manifestation identifier” that is listed is most
commonly the ISBN. This is an important data element, but I do wonder how
often users (including library staff) approach the library catalog with an ISBN in
hand (or head). The ISBN is, however, heavily utilized in automated processes,
such as duplicate detection and retrieval of cover images from online sources.
Because there is no definition of users in the document, it is not possible to know
whom the group had in mind for the various data elements, nor can we know if
some bibliographic attributes were specifically intended for automated processing.
©Karen Coyle CC-BY
DOES FRBR MEET FRBR’S OBJECTIVES? / 103
The statement of responsibility is among the elements that have a moderate
role for “find a manifestation.” This is not a heading in library data, and I am
confused by the assumptions the FRBR Study Group makes regarding the action
of “finding.” In fact, the report does not mention indexing, nor whether there is
even an assumption that there are headings. Yet the find action does imply that
some ability to search must exist, and the Final Report describes the elements of
moderate value for find as those “typically used as a secondary search term.” It
isn’t clear what “secondary search term” means, but presumably this is a term
that can be used to limit results, as with the use of limiting elements in many
catalogs by language, resource type, or other characteristic. The Study Group
clearly harbored some implicit assumptions about system capabilities, but what
these are is not made clear.
For manifestations, the only secondary element listed is “form of carrier.” This
is a typical secondary search term, but “form of expression” (i.e., whether it is
text, music, or film) is not considered of any value for the find task, even though
the advanced search of some systems includes the ability to limit by form.
Some of the moderate-value find elements are relationships: successors and
supplements are included, but not adaptations or transformations. Most of these
relationships are specific to serial publications. The FRBR-based catalog may not
lead you from a parody to the thing parodied, but it should allow you to make
the connection from a supplement to the thing it supplements. That serially
published items need to be connected is rather obvious, but it isn’t clear to me
why these relationships are attributed to the find function and not, for example,
identify or select.
The obtain task is not what I would have immediately expected. It would be
logical to think that obtain refers only to items held or offered by the library, and
that a primary element would have been the call number or some other identifier
that leads directly to the item itself. Instead, the obtain task includes “acquire an
entity through purchase, loan, etc.” This wording is now included in the functions
of the catalog listed in the 2009 IFLA International Cataloguing Principles. The
previous version of the Principles, dated 1961, echoed Cutter’s original functions
of the catalog by limiting those functions to “whether the library contains a particular book.” Thus the obtain task includes an expanded view of the role of the
catalog. For this reason, obtain relates both to manifestations and items, and the
key elements are those that define (or identify) the manifestation, including title,
statement of responsibility, publisher, and series. For example, place of publication
is considered of low value for identifying and selecting a manifestation, but is of
high value for obtaining it. Oddly, place of publication is also of low value for
selecting a work and an expression, even though, by the very definition of those
©Karen Coyle CC-BY
104 / CHAPTER EIGHT
entities, no place of publication is possible. The user tasks might have benefited
from an introductory discussion of the expected functions of the catalog. As it
is, those functions must be teased out by reading between the lines.
The answer to the question “Does FRBR provide functional requirements?”
is, for the most part, “no.” The derivation of entities and attributes is part of
the E-R analysis which dominates the FRBR Final Report, not of the analysis of
user tasks. Everything about the document revolves around the E-R analysis, the
entities, entity attributes, and the bibliographic relationships. When the user tasks
are analyzed, after the FRBR model has been fully developed in earlier sections
of the document, some of the attributes and relationships are shown to be of
little importance. No functional requirements are given for the entities themselves. To have provided a truly functional analysis would have required making
explicit some of the implicit assumptions that seemed to have been shared by the
members of the FRBR Study Group. Examples of these are assumptions about
system capabilities for search, and the goals of the catalog. The group obviously
believed its own statement that FRBR was technology neutral, even though they
used a specific technology, entity-relation analysis, to produce their results, and
that they had to assume, but never explain, certain types of search capabilities that
would make their attributes and relationships usable in some real implementation.
DOES FRBR PRODUCE A NATIONAL LEVEL
BIBLIOGRAPHIC RECORD?
The FRBR Study Group was also tasked with determining the bibliographic
elements of a basic level bibliographic record for national libraries (BLNB) based
on the results of their analysis. It does so in chapter 7 of the group’s final report.
This chapter reads like an afterthought, however, because it isn’t mentioned in
the rest of the text, and oddly its content has very little to do with any other
content in the report.
The BLNB defined in section 7.3 of the FRBR Final Report resembles ISBD
more than it does FRBR. It uses the ISBD concept of “areas,” such as “Title area”
and “Edition area,” which are not included in FRBR, and defines a short list of
bibliographic elements for each area. The terms work, expression, and manifestation
are not included in this section, nor does the BLNB make use of the attributes
that are defined in FRBR. Where FRBR has “title of the manifestation,” the
BLNB has “title proper (including number/name of part).” The BLNB refers to
“uniform titles,” a term that is not used in FRBR, and is not defined anywhere in
the report. Uniform title is also not included in ISBD, so it may not be known
to those who would need to use the BLNB. The BLNB also includes a lengthy
©Karen Coyle CC-BY
DOES FRBR MEET FRBR’S OBJECTIVES? / 105
“notes” area, even though no notes are included in the FRBR attributes. In fact,
the BLNB appears to have very little relationship to FRBR at all.
The relationship between FRBR and ISBD is unclear. The FRBR report does
cite ISBD as a source:
The basic elements of the model developed for the study—the entities, attributes,
and relationships—were derived from a logical analysis of the data that are typically
reflected in bibliographic records. The basic elements of the model developed
for the study—the entities, attributes, and relationships—were derived from a
logical analysis of the data that are typically reflected in bibliographic records.
The principal sources used in the analysis included the International Standard
Bibliographic Descriptions (ISBDs). (FRBR Final Report, 4)
Although throughout the FRBR study ISBD is cited as a source of attributes for
the entities, it is instructive to look briefly at a comparison of FRBR, ISBD, and
the Basic Level National Bibliographic Record in chapter 7 of the FRBR Final
Report. ISBD has ten areas and ninety-two attributes. FRBR has eighty-four
attributes, plus thirty-four bibliographic relationships. The numbers alone do
not provide a good comparison, however. FRBR does not include any note fields
amongst its listed attributes. ISBD, on the other hand, lists thirty-two note types.
The Basic Level National Bibliographic Record uses seven of the ISBD areas and
adds four more areas for access points (names, titles, subjects, and series). It has
fifty-three attributes, of which twenty-two are notes fields. Some of the notes
describe bibliographic relationships that are included in FRBR. For example,
the FRBR relationship “successor” in the BLNB becomes “note on edition and
bibliographic history—successor.”
The upshot is that the BLNB at least superficially resembles ISBD more than
it does FRBR, which makes its appearance in the FRBR Final Report particularly
jarring. It states that “the terms used to identify individual data elements correspond to the terms used in the International Standard Bibliographic Descriptions
(ISBDs) and the Guidelines for Authority and Reference Entries (GARE),” but
gives no explanation of why ISBD and not FRBR terms were used.
The introduction to chapter 7 states that the development of BLNB used
bibliographic concepts from FRBR work, expression, and manifestation, but
not item. In that section, the explanation for this decision is that it does not
include item because this is a national library record. However, it isn’t clear how
that follows unless this analysis specifically does not cover archives or rare book
materials, nor any other materials, like art and museum objects, for which itemlevel description is key.
©Karen Coyle CC-BY
106 / CHAPTER EIGHT
It is worth noting that FRBR itself resembles a minimum bibliographic description. Compared to MARC 21, FRBR is clearly a high-level core of bibliographic
data elements. FRBR has fewer than 100 attributes, while MARC 21 has nearly
1,500 unique data elements. Although FRBR defines a number of bibliographic
relationships, like “successor to” or “adaptation of,” there is only one relationship between a person and each FRBR Group 1 entity: creator (work), realizer
(expression), producer (manifestation), and owner (item). In comparison, the
list of relators that can be used for persons and corporate bodies in MARC 21
records has over 260 different terms.
The BLNB has not been instantiated by IFLA, even though this was a stated
objective of the FRBR Study Group’s work. It appears that this objective has
not be fulfilled.
DOES FRBR SERVE USER NEEDS?
For a study that was purported to be user-centric, the user’s absence is notable.
There is no analysis of users; no mention of how varied the library user base is;
no mention of children or elders or the disabled. Instead, to my mind, the FRBR
Final Report reads as a study by catalogers for catalogers. Even more specifically,
this is a study by catalogers at large research institutions. The list of participants
in the FRBR Study Group does not include anyone from a public library, nor
anyone from a non-Western country.
The FRBR document states:
The study makes no a priori assumptions about the bibliographic record itself,
either in terms of content or structure. It takes a user-focused approach to
analyzing data requirements insofar as it endeavours to define in a systematic
way what it is that the user expects to find information about in a bibliographic
record and how that information is used. (FRBR Final Report, 3)
It cannot be coincidence that this study describes a bibliographic description
that looks incredibly like the one that already existed at the time. As Le Boeuf
says in the introduction to his 2005 book on FRBR: “FRBR models what we
do, not what we should do.” To say that the study began with a bibliographic
blank slate, yet ended up describing the precise data elements that are in bibliographic standards like ISBD, defies credibility. The mere presence of elements
like “statement of responsibility” shows that the needs of users did not drive the
development of FRBR, because it’s pretty clear that no nonlibrarian user would
©Karen Coyle CC-BY
DOES FRBR MEET FRBR’S OBJECTIVES? / 107
call for a statement of responsibility. In fact, the FRBR Final Report states that
“the attributes defined for the study were derived from a logical analysis of the
data that are typically reflected in bibliographic records.” This is circular reasoning,
which should explain why some entities seem under-defined or underutilized.
For example, there is a place entity for subject, but no place entity that can be
used for place of publication, because the latter is not given authority control in
current cataloging:
Inasmuch as the model also defines place as an entity it would have been possible
to define an additional relationship linking the entity place either directly to the
manifestation or indirectly through the entities person and corporate body which
in turn are linked through the production relationship to the manifestation. To
produce a fully developed data model further definition of that kind would be
appropriate. But for the purposes of this study it was deemed unnecessary to
have the conceptual model reflect all such possibilities. (FRBR Final Report, 31)
How the FRBR Study Group addresses the question of what users need (or even
what they want, which is not necessarily the same) is one that I find particularly
disturbing. Not only were users not consulted about any aspect of this but the
report provides no analysis of existing research.
So how did the FRBR Study Group study users? According to Olivia Madison:
The FRBR Terms of Reference mandated that the study take into account the
needs of a broad range of users and how they use bibliographic records. . . . One
obvious option was to query, using a systematic methodology, a broad range
of users, and draw conclusions from this analysis. Another option was to use
our collective knowledge of the various types of users from the working group
membership and commentators, as well as to draw upon experts in the fields to
provide necessary user perspectives and conclusions. The Study Group decided
in favor of the latter approach. (Madison 2005, 28–29)
My note to myself on that page reads: “We don’t need no stinkin’ user surveys.”
The FRBR Final Report implied that some study was done of the literature:
The assessment of importance of each attribute or relationship to a given user
task that is reflected in the tables was based in large part on the knowledge
and experience of the study group members and consultants, supplemented by
evidence in the library science literature gathered from empirical research, as
©Karen Coyle CC-BY
108 / CHAPTER EIGHT
well as by assessments made by several experts outside the study group. (FRBR
Final Report, 83)
In other words, the study of user needs was done without studying users. But
in fact the Terms of Reference that gave the group its charge also says very little
about users and does not require the group to study them. In addition, the
development of the user tasks was not dictated by the Terms of Reference.
The document itself is not structured around the user tasks, and those tasks
are defined only at a very high level before the document begins its description
of the three groups of entities of bibliographic control. Within the document
sections that describe the entities, the attributes, and the relationships between
entities, there is no discussion of how the user tasks inform these aspects of the
model. Although there is a section devoted to the relationship between the user
tasks and the attributes, there is no discussion of how different attributes help users
make their decisions, only a statement that they do provide information for users.
For example, in section 6 on user tasks, the identify task is described in this way:
1. the attribute by definition serves to identify the entity (e.g., manifestation
identifier, item identifier);
2. the attribute or relationship forms part of the minimal set of attributes and
relationships that for the majority of cases will serve, in the absence of a unique
identifier, to differentiate entities that have a number of common characteristics
(e.g., the minimal set of attributes sufficient to differentiate one manifestation
from another in the majority of cases comprises title, statement of responsibility, edition/issue designation, publisher/distributor, date of publication/
distribution, series statement, and form of carrier). (FRBR Final Report, 85)
This leads one to wonder just how the user is defined. The identifiers in point
one are generally for machine processing, not for humans, although occasionally a human may come to the catalog in possession of a relevant identifier, like
an ISBN. Are machines included in the group’s concept of users? On point
two, unless the user is a trained cataloger, this is a completely unfounded set of
assumptions. How do users actually identify manifestations? Do nonlibrarian
users pay any attention to the statement of responsibility? Do they understand
the series statement? In fact, do they even see these data elements while making
their choice? Many systems only display these in a full display, and users must
make their choice from the results list with brief bibliographic displays.
The main differences between FRBR and our current bibliographic records are
the division of the bibliographic description into four entities, and the potential
©Karen Coyle CC-BY
DOES FRBR MEET FRBR’S OBJECTIVES? / 109
for the creation of linking relationships between bibliographic entities. Although
the latter holds promise, there does not yet exist a system that would provide
evidence of user interaction with these relationships. There have been some studies
providing system displays that use the Group 1 entities to cluster retrievals, and
these have shown that the presentation of clusters of works is often preferred
to the list of redundant editions that catalogs generally provide. This begs the
question, however, of whether the FRBR Group 1 coincides with the user view
of the bibliographic world.
Two studies by Jan Pisanski and Maja Žumer (2010) on residents of the
Slovenian capital of Ljubljana investigated this question of whether the FRBR
Study Group’s view of Group 1 entities is consistent with how users view the
bibliographic universe. They titled their studies “Mental Models of the Bibliographic Universe.” The studies used card sets that presented library users with
bibliographic information consistent with the entities of Group 1. For example,
using Dan Brown’s Da Vinci Code, there were cards for:
``
``
``
``
Dan Brown’s Da Vinci Code
Dan Brown’s Da Vinci Code, published by Doubleday in New York in 2003
The Slovenian translation of The Da Vinci Code by Nataša Müller
The movie of The Da Vinci Code starring Tom Hanks, in English with
Slovenian subtitles
and others. Note that the users were not given individual Group 1 entities, but
instead were presented with either a work; a work and an expression; a work,
expression, and manifestation; or the entire Group 1 from work to item. This
is consistent with the conceptual view of Group 1, although not its entity-relation view.
The users were asked to place these cards in order from the most abstract
to the most concrete. This instruction was difficult for many users because
they hadn’t previously applied the concepts of “abstract” and “concrete” to
bibliographic items.
In a second study, users were asked whether, of a pair of items, either item
was a reasonable substitute for the other.
The authors made some key discoveries:
1.Users have many different views of the bibliographic universe, but where
users had a common view, it generally was a FRBR-like view of the
“progression” from a general concept (work) to individual publications
(manifestation) and lastly to specific items (such as signed copies).
©Karen Coyle CC-BY
110 / CHAPTER EIGHT
2.What mattered most to users was the language of the text, the form (book
versus film), and the contents (illustrations versus no illustrations).
3.Users have a strong sense of “original work” which was the item that they
placed at the “top” of the hierarchy, and for them took the place of the
FRBR abstraction, work.
4.Users seek items at the FRBR level of expression for the most part, and
consider the related manifestations to be acceptable substitutes for each
other.
These results confirm some of the assumptions built into FRBR and into bibliographic cataloging rules, but they also reveal interesting variations. First, it
should not be surprising that the results of the card sorting exercise did not turn
out a single view of the tested items. Within the cataloging community there is
debate over the exact definitions of work and expression (less so of manifestation),
in spite of there being a shared knowledge of current cataloging concepts within
that community. Nonlibrarian users (and one could extend that to librarians who
have not been trained as catalogers) should be expected to bring a wide range of
views to the table. The test group, though, was given strong hints through the
experiment, which represented cataloging that had already surfaced a particular
view of the bibliographic items based on FRBR’s Group 1.
The discovery that people have a sense of the “original work” is quite sensible.
There is nothing in FRBR, nor in the RDA cataloging rules based on FRBR,
which directly addresses the question of “original work.” Yet there is obviously
a progression from the creative form that comes directly off the efforts of the
inventor of the work and all versions that follow it. For Patrick Wilson’s concept
of bibliographic families, the original work would be the family’s progenitor. In
the case of translations of texts, there is an implicit concept of an original form
that is translated to some other language. This progression is less strong in other
areas, such as in music, where there may not be a recording that one could think
of as an original performance of the piece. The separation between film and text
can be clear, as in a film version of Da Vinci Code, but it is less clear for the Star
Trek series of books that were written after the television series aired.
The conclusion that users seek language materials at the expression level is
quite logical, and this coincides with Shoici Taniguchi’s theory of an expression-dominant model for texts. The work is an abstraction without expression,
and in practice the work would represent all language versions of an oft-translated
work, or all editions of a work that had been updated, such as a yearly almanac.
Most users have a language preference for reading, and therefore the work set
©Karen Coyle CC-BY
DOES FRBR MEET FRBR’S OBJECTIVES? / 111
that includes all languages of the text would not be useful. In the case of different revised editions, such as with textbooks or reference books (which was not
tested for in the Pisanski and Žumer studies), it would be unusual for a user to
be seeking all editions, and even more unusual for a user to be seeking something
other than the most recent edition available.
For the user at the catalog, the FRBR user task find may not visibly change if
FRBR is adopted in the future. First, we must assume that users will not be aware
of the Group 1 structure, but will search as they do today, which means either
by keyword, which pools words from all of the searchable fields in the record, or
by creator or contributor, title, or subject.
I am assuming that these searches will continue to work in this way, with the
user not being required to know what Group 1 entity the search should go against.
This would be consistent with how searching is done in catalogs that are not
organized around the FRBR entities. For example, in most online catalogs, the
title index includes all or most title elements from the record; therefore the user
searches the work title and the manifestation title with one search along with any
alternate titles. The same is true of the various creators and contributors to the
resource, who are often searched together in the same index, whether authors,
illustrators, composers, or librettists. This practice will probably remain, even
though in FRBR primary creators are linked to the FRBR work and some secondary creators such as translators and illustrators, are linked to the expression
or even the manifestation.
Where changes are anticipated in the catalog, however, are in the user displays.
There is a general assumption that users will not see a manifestation-level display
as they do today, but will be given a view that takes advantage of FRBR Group 1
as a way to gather all versions of the work together in a new kind of collocation
using the primary author and the work title. However, there may need to be
decisions based on the language of the catalog or of the user. For example, the
work title Война и миръ (War and Peace) may not be useful in a catalog aimed
at English-language speakers, yet that is the correct work title as defined in the
cataloging rules. Whether that title should be displayed to the user or not is a
question for user-interface designers. The expression entity in catalogs today is
often represented by a facet that allows a user to narrow the retrieved set by language. Because FRBR is only a conceptual model, how the FRBR entities might
be used in future catalogs is not specified, and the document does not provide
any suggested display forms.
For some materials, the FRBR user view may be a vast improvement over the
record-level view of today’s online catalogs. The “Scherzo” project at Indiana
©Karen Coyle CC-BY
112 / CHAPTER EIGHT
University, as reported by Handesty et al. (2012), developed a FRBRized catalog
of music materials and did comparative user testing between the FRBRized and
the traditional catalog. Although some of the results were mixed, they concluded
that users preferred the FRBRized catalog.
The benefits of an FRBRized display are often used as an argument for moving
to FRBR as the basis for cataloging. In fact, the FRBRized displays that have
been tested have mainly been built from bibliographic data that was not cataloged
according to FRBR principles or stored digitally as FRBR entities. WorldCat is
developing a work-based display from its pre-FRBR bibliographic records, as
did the library vendor VTLS. Outside of the library world, bibliographic databases like LibraryThing and OpenLibrary have developed work-based displays
from data that was created without the FRBR entities as concepts. A FRBRized
display that used data that was originally designed around FRBR principles may
be superior, but there may be ways to take advantage of some FRBR concepts
without having to entirely re-engineer library systems.
There is one key user-related concept that gets no mention in the FRBR Final
Report, and that is “collocation.” Today’s catalog record is a single metadata
container that carries all of the entities described by the FRBR Final Report. The
primary focus of this record is what FRBR would consider to be a manifestation,
and in most systems the dataset returned from a search is a group of manifestation-focused records presented in some order (ranked, by date, or alphabetical).
With our bibliographic universe organized as defined in FRBR Group 1, however,
there are new options relating both to retrieval and to display. FRBR has the
potential to restore the context of bibliographic relationships that existed in the
collocation function of the card catalog. This loss of collocation has been a source
of frustration for catalogers, as I have learned from many conversations about
library systems. Martha Yee described online catalogs as mere “finding lists,”
and suggested that “FRBRization” could produce a list of works, expressions,
and manifestations related to a user’s search so that users could make use of that
bibliographic context to select resources of interest (Yee 2005).
Perhaps it is deemed to be obvious from the inclusion of bibliographic relationships that a major function of FRBR is to reintroduce the collocation function
that was lost to the online catalog, when discovery through the linear order of the
physical catalog was replaced by database-based retrieval of individual bibliographic
records. Collocation has always been a surfacing of relationships between bibliographic items, although in the physical catalog the relationships could not be
made explicit. It was left to the user to intuit the meaning of the physical proximity
between cards, if there was one at all. The separation of bibliographic data into
©Karen Coyle CC-BY
DOES FRBR MEET FRBR’S OBJECTIVES? / 113
entities is emphasized in the FRBR Final Report, although the real potential for
user service comes from the ability to not only collocate related bibliographic
items, but to make explicit what that relationship is. Although the Pisanski and
Žumer studies look mainly at how users organize the elements of FRBR Group
1, they show, perhaps inadvertently, that users are cognizant of bibliographic
relationships like translations and adaptations, even though they may not have a
conscious awareness of those relationships and what they might mean.
The cataloging rules based on FRBR attempt to create a shared concept
of the entities and how they are described so that libraries can share equivalent bibliographic data. What the Pisanski and Žumer studies tell us is that we
cannot assume that untrained users will immediately understand how libraries
have divided up the bibliographic universe because it may differ from their own
underlying assumptions, although with the language materials that they studied
there is much overlap between the user view and the FRBR Group 1 concepts.
The advantage of consistency is that it provides a platform that can be learned;
whether the general public will benefit from that, and who within that general
public will benefit, is not known.
DOES FRBR PROMOTE CATALOGING EFFICIENCY?
One of the primary reasons behind the development of FRBR was to create a
more efficient environment for cataloging. This came out of the 1990 Stockholm
IFLA meeting. The combination of lowered budgets and increased information
resource output creates a perfect storm for libraries, and especially their technical
services departments. So it makes sense to ask whether FRBR is likely to increase
cataloger efficiency. This is one of the goals behind the FRBR Group 1, because it
makes it possible to share cataloging at the work and expression levels, and not to
repeat data elements for works and expressions for each manifestation. In a system
with separately stored works and expressions that are linked to manifestations,
the cataloger would not have to repeat the work and expression information,
but only provide a link from the manifestation to the expression. This is seen
as a savings of time for catalogers, as well as potentially a point of efficiency for
systems. Although some libraries have been cataloging under the RDA cataloging rules for over two years now, existing systems do not take advantage of the
entity-relation structure of FRBR, so they have not gained experience with the
ability to share separate descriptions of works and expressions.
One of the difficulties in answering this question is that libraries are not a
homogeneous group, and cataloging is not the same activity throughout the
©Karen Coyle CC-BY
114 / CHAPTER EIGHT
library world. In particular, national libraries and large research libraries probably
do the greatest share of original cataloging, and the most detailed cataloging. A
considerable number of smaller libraries only copy catalog data created by others
for common published materials. Others purchase a great deal of their cataloging
from the library vendors who supply their books and other materials. Within
the libraries that rely heavily on copy cataloging, there is variation in how much
editing is done of the cataloging data to customize it for the individual library.
How much one gains from the division of bibliographic data into entities
depends on how often one can reuse entities that already exist. That is not easy
to assess. More than one study has been done to try to quantify the proportion
of works that are in defined bibliographic relationships with other works. Richard
Smiraglia (2001) studied a small number of general academic libraries as well
as libraries specializing in music and theology. Not surprisingly, the specialist
libraries showed different patterns of bibliographic family membership. Sherry
Vellucci’s 1997 study of a music library showed that over 85 percent of the
holdings were in a bibliographic family that could involve some sharing of data.
In a general academic library the percentage may be as much as 50 percent. In
OCLC’s WorldCat database, which is more varied than most catalogs due to its
international nature, about 30 percent of manifestations are in a bibliographic
family, as per the study done by Brian Lavoie in 2001, as well as more recent
statistics comparing numbers of works with total records.
Smiraglia also shows that having family members is a function of time: older
works are more likely to be in families. A rather obvious, but worthwhile, point
to prove is that that works go out of print and are republished and repurchased;
and that derivative works occur only after time has shown that the work is worthy
of further study and derivation. In the theology library collections studied by
Smiraglia the mean age of works was 114 and 125 years, while general collections
had a mean age of 32 years. Collections in non-research libraries are probably
much younger than those studied by Smiraglia. Smiraglia’s studies also showed that
form and genre were not predictors of family relationships in the libraries studied.
One can conclude that the nature of the material being collected has a great
influence on the incidence and size of bibliographic families. None of the studies,
however, looked at nonacademic libraries, whose collections are not analogous
to academic libraries, nor did they analyze topic areas other than theology and
music. With a reminder that these are studies of library holdings and not of
publication patterns, one could expect that where scientific works are updated
and republished, older editions might not remain on the shelves, and therefore
are not included in the catalog. Small and medium public libraries may be
©Karen Coyle CC-BY
DOES FRBR MEET FRBR’S OBJECTIVES? / 115
expected to have fewer family members due to the exigencies of limited shelf
space and the mission to serve the reading, listening, and viewing public more
than research activities.
In the previously mentioned analysis of manifestations and works in WorldCat,
published by Brian Lavoie and colleagues in 2001, yielded this information:
Number of manifestations (records)
Number of works
46,767,913
32,000,000
Average number of works per manifestation
1.5
Number of works with only a single manifestation
78%
Number of works with seven manifestations or less
99%
Number of works with more than twenty manifestations
1%
Because of the nature of the WorldCat database, which represents the holdings of
many tens of thousands of libraries worldwide, these figures cannot be extrapolated
to be meaningful for a single library. On the other hand, because of its breadth,
WorldCat gives us a reasonable picture of publishing patterns over the combined
universe of library holdings. This tells us that a large number of published items
appear in only one version, and therefore the manifestation-expression-work
ratio for those items is 1:1:1. Using the 2001 OCLC figures, 78 percent of the
items in WorldCat are single manifestations of this type. Only 1 percent of the
identified works have been produced in a large number of manifestations. The key
question now becomes: What do these figures mean for the cataloger workflow,
and would a cataloging system that makes work and expression data available
for reuse result in greater efficiencies than we have with a cataloging model that
is manifestation-centric?
To answer this question we would have to know how many previously described
works a cataloger is likely to encounter when processing new items today. The
figures that we have only reflect the cumulated past, not the present or future publishing output that catalogers confront. Regardless of the number of expressions
and works in a bibliographic family, every new manifestation must be described.
We should be able to conclude that there is no change in terms of cataloging
efficiency in the description of manifestations. The new efficiencies would be
found in the aspects of the bibliographic record that describe the expression
and the work. There are three possible situations that a cataloger doing original
cataloging may encounter:
©Karen Coyle CC-BY
116 / CHAPTER EIGHT
1.
A manifestation that represents a new expression and a new work. This would
presumably be the case for newly published first editions. The cataloger
creates the necessary data elements for the manifestation, the expression,
and the work.
2.
A manifestation that represents a new expression of an existing work. The
cataloger creates the necessary data elements for the manifestation, the
new expression, and links to the existing work.
3.
A manifestation that represents an existing expression (and thus its related
work).The cataloger creates the data describing the manifestation, and
links the manifestation to the existing expression.
All of these represent the activities of original cataloging. It does not appear that
the activity of a cataloger who copies catalog data created by others is greatly
changed from what is done today, which is to locate an existing bibliographic
record for the manifestation in hand and to utilize that data for the local catalog.
Whether or not FRBR is efficient for catalogers, and whether the existence of
FRBR work data saves time, depends entirely on how often catalogers encounter
situations two and three, above. The WorldCat study only yields statistics for
manifestations and works, not expressions, and that study presents a snapshot
of the WorldCat database, which may not represent the bibliographic situations
encountered on a daily basis by today’s cataloger. Obviously, the majority of
older published works have already been described, so what interests us is how
often those are reissued as new manifestations, and whether today’s publishing
patterns are producing more or fewer instances of reuse of expressions and works.
In addition, catalogers in different types of libraries (for example, medium-sized
public libraries versus large research libraries) will encounter a different pattern
of publication types. None of the above can be applied to catalogers of unique
archival materials that do not benefit from the sharing of bibliographic description.
The gain in cataloger efficiency may be greater for some materials, like music
recordings, which have a high degree of repetition of works.
It is necessary here to reflect on the effect of the changing nature of library
materials from solely physical to digital. One of the motivations for reviewing
library cataloging, beginning in the early 1990s, was the recognition that libraries now often receive the same content in more than one physical format. This
was prescient at the time because only a few digital formats were commonly
available. Today, with the increase in the number of e-books and e-journals, it
is quite common for a library to offer both a physical and a digital copy of the
same resource, and even multiple forms of the digital version. The cataloging
©Karen Coyle CC-BY
DOES FRBR MEET FRBR’S OBJECTIVES? / 117
rules regard these as different manifestations, but some libraries have been loath
to confront users with multiple records for the same content in different forms.
This was the problem called “multiple versions” that was much discussed in the
1990s. Many libraries have “solved” this problem by providing links to digital
versions on the records for the hard copy version, thus providing the user with a
single point of entry for either format. This violates the cataloging rules, however,
because a manifestation is defined by its physical format. Using FRBR Group 1
as separate described entities means that all of the multiple physical versions can
be associated with a single expression that is subordinate to a single work. This
could allow grouping of multiple versions under work or expression in library
catalogs, while still adhering to the emphasis on physical formats that motivates
the cataloging rules. Of course, there could be many ways to achieve this kind of
grouping; the FRBR entity-relation structure is only one of those.
We can conclude that we need much more data before we can answer the
question of whether FRBR Group 1 actually saves time for some catalogers. In
addition, whether the entire FRBR entity-relation model results in greater efficiencies may depend more on systems designs than on anything specific in the
FRBR model. In current cataloging, name and subject authorities are already
analogous to FRBR’s Groups 2 and 3, and these are shared as records representing
specific members of those entity types. While indeed the FRBR entity-relation
model appears to promote efficiency in cataloging, there is as yet no concrete
evidence that it will.
DOES FRBR AID DATA SHARING?
Another question, however, is how well FRBR works for sharing of bibliographic
data between library systems. In the early days of modern cataloging, beginning
in the middle of the nineteenth century, each library was an entirely separate
entity, and the bibliographic universe was limited to the holdings of that library.
Today library data is shared internationally.
There is a tension between the shared bibliographic universe and navigation
within the catalog of a single library. This conflict could become even greater with
the application of FRBR relationships as machine-actionable links. In today’s bibliographic records, many of the relationships that we assume will be links in FRBR
are presented as notes (“Translation from the German.” “Based on the book by
. . .”) or as assumed from edition statements (“3rd ed., revised and enlarged”).
These notes do not provide useful linking, but they do provide information for
the user even when the library does not hold the item referred to in the note.
©Karen Coyle CC-BY
118 / CHAPTER EIGHT
Links alone will not provide that information for catalog users in cases where the
links do not resolve to another bibliographic description in the library’s catalog.
Ultimately, users need both information and direct links where available. The
FRBR entity-relation model does not address implementation issues of this nature,
so we should assume that FRBR is a starting point upon which development may
take place, even if that development results in a modified model.
The FRBR relationships are universally bibliographic in nature; they definitely
go beyond the inventory role of the library catalog. In a large research library
the holdings may be extensive enough that the disconnect between bibliography
and inventory remains at tolerable levels. Even so, what is the library to do with
machine-actionable relationships to items it does not own?
There is an ideal bibliographic universe where every work ever produced is
placed in its proper place, and the whole can be navigated intelligently. There
is no real version of that bibliographic universe, however. Nonetheless there is
a desire to use today’s technology to create the “library of everything,” though
“everything” has a lot of ragged edges. It also has a lot of individual libraries with
different populations. FRBR as written in the Final Report appears to address an
abstract bibliographic universe, which may or may not coincide with the reality of
individual libraries. The tension between bibliography and the catalog will have
to be resolved. FRBR operates at an abstract level that needs to be reinterpreted
for the many thousands of use cases that exist in libraries, and still allow them
to share cataloging copy.
This tells me that we need to define the relationship between the library catalog
and the bibliographic universe. I can imagine a multitiered design that can move
a user from the inventory of the single library to related bibliographic items that
are nearby or otherwise available (e.g., for purchase). I can also imagine a linked
bibliographic space that is not dependent on library holdings, but that is about
the relationships between bibliographic items. This would need to go beyond
FRBR, because some of the more useful relationships (e.g., “cites” or “quotes”)
are not included there. Connecting this “cloud” of bibliographic relationships to
actual availability in specific libraries or bookstores would be essential. Whether
we can develop the technology to realize this vision remains to be seen.
IS FRBR FORMAT NEUTRAL?
A model like FRBR needs to accommodate all of the types of resources that
will be included in the library catalog. This is a difficult requirement because
libraries and archives manage a wide variety of materials whose bibliographic
©Karen Coyle CC-BY
DOES FRBR MEET FRBR’S OBJECTIVES? / 119
natures are far from uniform. Although the FRBR Study Group stated that
FRBR was designed to describe a model that could be applied to any type of
bibliographic data, numerous articles have discussed how the conceptual model
would need to be reinterpreted for specific resources types. Some of those types
are presented below.
Serials cataloging deals with a particularly complex set of relationships: the
whole/part relationships between the resources that users generally seek, the
articles, and the continuing resource that is the serial itself. There is also the
very difficult question of what is the serial work, and when does a serial become
a new serial?
There are relationships defined in FRBR that are suited to serial publications,
such as Successor and Supplement, and properties such as Sequencing Pattern
and Expected Regularity of Issue. As is often the case, however, the difficulties
come in the interpretation of the Group 1 entities.
Paul Weiss and Steve Shadle (2007) look at “FRBR in the Real World.” They
point out that it isn’t obvious that a serial is itself a work in the FRBR sense of
that term, that is, “a distinct intellectual or artistic creation.” It is easy to see serials
as aggregations of works, but the serial itself appears to lack that distinctness that
FRBR requires of works. As we’ll see in a later section, aggregations of works
present a number of problems in FRBR, including the fact that each aggregated
work itself is defined as having the full complement of Group 1 entities, which
means that whole/part relationships between the larger entity, the serial, and
the part entities, the articles, can be quite complex. This complexity is only the
beginning, because serials publishing provides some stunningly difficult examples, such as serials within serials, supplements to serials, and a wide variety of
enumeration patterns.
Kristin Antelman (2004) also addresses the question of seriality and the work.
She notes that the main identifiers for works, both in FRBR and in earlier cataloging theory, such as author and title, are not strong identifiers for serials. Serials
often do not have “authors” per se, and because their titles so often change, one
journal of serials cataloging is itself titled Title Varies. Serials also have been the
main resource type to be provided in digital format as well as print, with the
result that libraries often have both print and digital subscriptions to the same
content but packaged differently. Antelman recommends “asserting bibliographic
control over a higher level of abstraction than has been our practice,” and putting
©Karen Coyle CC-BY
120 / CHAPTER EIGHT
a “greater emphasis on relationships between abstract entities and less on the
identification of the physical item.”
Ed Jones, in his article on “The FRBR Model as Applied to Continuing
Resources” (2005), points out that the boundaries between work, expression, and
manifestation in continuing resources differ from that of other materials because
changes in key elements, such as the title, can take place in one expression or
one manifestation but not in the others. Because serials cataloging leans heavily
on titles for the identity of serials, this poses a conflict between the title-based
definition of work in serials cataloging and the definition of work in FRBR.
Jones’s argument is not that the bibliographic entities in FRBR are problematic,
but that they are not in accord with the rules for serials cataloging. Either FRBR
or the serials definition of the work as being title-based would need to change.
This sounds radical, but it should not be surprising that a new conceptual model
could lead to changes in cataloging practice, or that cataloging practice should
inform modifications to a conceptual model.
One possible positive outcome of FRBR or a FRBR-like linking model is that
it may be easier to link articles and the journals they are published in. Laura Krier
(2012) sees advantages in the use of linked data for serials. Krier’s analysis isn’t
specific to FRBR, but FRBR relationships are one possible way that serials linked
data could be achieved.
Music is an interesting cataloging case even as presented in the FRBR Final
Report itself. The relationship of work and expression in music often takes place
as a performance, and individual works can have a very large number of expressions. Yet, music cataloging (primarily for classical music) makes more use of
work titles than any other specialized type, which means that the concept of the
work is strong among librarians and probably also among users.
Raymond Schmidt, however, takes a look at jazz and improvisation in his
“Composing in Real Time: Jazz Performances as ‘Works’ in the FRBR Model”
(2012). If each performance of a jazz musical piece brings in new sounds, the
difference between a work and an expression is blurred, and there are then
questions of whether or when a performance creates a new work. Musicians and
listeners recognize familiar tunes, and might consider two performances to be
of the same work. Yet the musicians who improvise performances or engage in
sampling of the works of others often consider themselves to be creating new
works even though a recognized “air” is present.
©Karen Coyle CC-BY
DOES FRBR MEET FRBR’S OBJECTIVES? / 121
The Variations Digital Music Library 2 Project at Indiana University, as reported
by Jenn Riley at the 2008 International Society for Music Information Retrieval
conference, analyzed the FRBR attributes for each WEMI entity and had a number
of modifications that were required to express the music materials. That analysis
concluded that the FRBR concept of a work may be suitable for the “canon of
Western art music, where composition takes precedence over any given performance of it.” However, they found that there were additions needed to FRBR
work for jazz, World, and non-Western music, which when music is not in written
form are the musical equivalent of an oral tradition. To describe these musical
forms correctly, the FRBR work would need the additional attributes of language
and place of composition. They also suggested that musical key was needed for
both works and expressions. (In FRBR it is only an attribute of work.)
Music publishing also presents some difficulties. Many works are issued in a
kind of “bound with” relationship with other works. Music manifestations have
the added complication of accompanying materials, such as liner notes, which
need a language designation in the manifestation, but not in the expression.
Riley provided this comparison of FRBR and the Variations project’s versions
two and three:
VARIATIONS 2/3 ENTITY
FRBR GROUP 1 ENTITY
Work (more concrete than FRBR Work)
Work
Instantiation (can only appear on one Container)
Expression
Container (includes some copy-specific data)
Manifestation
Media Object (defined as a digital file)
Item
This illustrates that the FRBR work may be an organizing concept rather than
a fixed “thing” that organizes at a different level of abstraction within different
contexts. In classical music, a performance based on the same score could be
considered the same work, while in jazz or hip-hop a performance might indeed
be the creation of a new work.
Martha Yee, a cataloger at the University of California at Los Angeles film
archives, found that the FRBR definition of work accords with that of the moving
images cataloging community, which is that the translation of a textual or musical
work to film always creates a new work. However, this view illustrates a distinct
©Karen Coyle CC-BY
122 / CHAPTER EIGHT
difference between film catalogers and other catalogers—most notability those
working with music materials:
Film catalogers consider a film of a performed work to be a new work related
to the previously existing text for the performed work. Thus a film cataloger
considers Bergman’s Magic Flute to be a new film work, not an edition of
Mozart’s opera. Music catalogers disagree; to them, Bergman’s film is still primarily Mozart’s work; that is, not a new work, but a new expression of Mozart’s
work. (Yee 2007, 124)
As Yee describes FRBR, “Change from any other GMD [General Material Designation or physical type] . . . into the moving image GMD . . . creates a new work
by [the] FRBR definition” (Yee 2007, 121). This contrasts with the music cataloging view that a new performance or recording is a new expression. In a library
catalog containing both music and film materials, these two views cannot coexist.
James M. Turner and Abby A. Goodrum (2009) address the many types of
editing that can be performed on moving pictures, including removing content
for reasons of audience (profanity, nudity) or to make the film fit the time allotted. These, according to the authors, are treated as new manifestations, where
in the text world these would undoubtedly be considered new expressions. At
the same time, there is interest in what Turner and Goodrum call “frame-level”
access to film, with examples like the Zapruder film of the Kennedy assassination
and various films of the 9/11 events. These are spontaneous creations, sometimes
even accidental, and so the designation of a creative work component may be
difficult. However, should such unplanned footage be included in a documentary,
its nature appears to change. Three people standing side-by-side at a scene may
take nearly identical photos or films of an event. How many works are there?
There are also the multitude of similar scenes shot for a commercial film which
may or may not be included in the film, and yet may be reconstructed later into
a new version of the film as the “director’s cut.”
Although library standards attempt to cover a wide range of library and museum
materials, there is a significant distance between the description of published materials and the treatment of unique materials, such as works of art. The cultural object
community has its own cataloging rules, Cataloging Cultural Objects (CCO),
which differs considerably from the rules for printed and published materials.
©Karen Coyle CC-BY
DOES FRBR MEET FRBR’S OBJECTIVES? / 123
In an analysis of FRBR in relation to art materials, Baca and Clarke (2007)
point out that the separation of work, expression, manifestation, and item is not
applicable for many of the materials they catalog. The CCO rules define a work
as a human-made object, meant to be perceived through the sense of sight. They
go on to say
Because of this sense of physicality, work, expression, manifestation and item
are one and the same for many cultural works, which are embodied in a single
material object and not in an abstract entity. (Baca, 104; emphasis in the original)
The CCO entity diagram has a central work, with authority-like entities for names,
geographical entities, concepts, and subjects. The central work entity encompasses
all of what in FRBR makes up Group 1.
The authors concede that FRBR Group 1 entities may be suitable at times
to conceptual art and artwork that takes place over time and in different forms,
where the artist’s concept and the execution of the art are not a single action.
These do not always break down into the WEMI categories of FRBR, however. It
appears that the description of artworks may need to make use of a variable interpretation of the Group 1 entities, as necessitated by different types of artworks.
A simple example is in the role of color in artworks. In FRBR, color (“colour”)
is an attribute of the manifestation, and is defined as “used in the production of
an image” (FRBR Final Report, 46). For some images, such as abstract paintings,
color is the content of the work and defines the work. One need only look at the
paintings that are variations on the theme of “white on white” to understand
that color can be the idea of the work.
The CCO view has equivalents to Group 2 entities, but it differs in how it
treats what in FRBR are the Group 3 subject entities. CCO uses two subject
entities, geographic place and concept, but does not list object or event as primary
entities. The subject authority in CCO is used to depict the subject matter of the
artwork, that is, what it is about. The concept authority is a thesaurus of terms
used in description, perhaps similar to the various controlled descriptive terms
developed for MARC records or for the RDA cataloging code. This thesaurus
includes terms for objects, materials, activities, agents, physical attributes, and time
periods. In the final analysis, much of the same descriptive territory is covered
although the organization of the elements is somewhat different.
©Karen Coyle CC-BY
124 / CHAPTER EIGHT
Because maps are often published in editions, the use of the FRBR expression
entity promises to help organize these materials. Two articles that focus on maps
and FRBR, one by Ruth Kalf and the other by Scott R. McEathron, illustrate the
use of FRBR to describe maps of the sixteenth through eighteenth centuries. This
was a time when maps were uncommon and the few that existed were copied and
reproduced in different languages. Those maps coincide relatively well with the
FRBR document-oriented model. For modern machine-rendered maps, which
can be produced easily in different sizes and scales, Kalf questions whether scale,
which seems appropriate at a FRBR expression level for the older maps, could
not be described at a manifestation level. This exemplifies the need for flexible
definitions of the FRBR entities even within a single cataloging specialty. In this
case, reproduction technology has changed how the resource is created and
realized, and a change in map scale is now more closely analogous to converting
a document from one word processing format to another.
Summary
It is quite possible that few of those among us who consider themselves to be
reasonably familiar with FRBR are actually aware of the original goals of the
project as stated in the Terms of Reference, even though these are mentioned in
the FRBR Final Report. It is also possible that many are not even aware of the
seventh section of the Final Report that defines a minimal level record, which
was the primary motivation for the creation of the FRBR Study Group.
The FRBR Final Report provides a revolutionary view of bibliographic data
as conceived in the service of the library catalog. It will probably be viewed by
future generations of librarians as a watershed moment in the history of the
library catalog. The gap, however, between the original goals and the FRBR
Final Report is great. The FRBR Study Group went beyond the originally stated
goals in an attempt to solve problems that had been bubbling to the surface of
cataloging practice over at least the previous half century. The task that arose
from the 1990 Stockholm meeting became the precipitating factor for a general
review of descriptive cataloging and its purpose.
That the FRBR Final Report overstepped the original goals is of historical
interest, and should help us understand that the context for FRBR is not in a
single meeting in 1990 but in all that came before it.
©Karen Coyle CC-BY
N IN E
SOME ISSUES
T H AT A R I S E
IN FRBR
T
here are some fundamental problem areas in FRBR as I read the document.
There have been hints some of these in the preceding description of the
FRBR entities: marked differences between the underlying structure of the
three groups and a lack of cohesion within the groups themselves (e.g., there are
no shared qualities among group members). Some of these issues just don’t feel
quite right; others pose problems in implementation. Among the truly difficult
problems are those of inheritance and hierarchy, of the disjoint nature of the Group
1 entities, and how bibliographic aggregates fit (or do not fit) into the FRBR model.
INHERITANCE AND HIERARCHY
It is a common assumption that the four entities in Group 1 represent a hierarchy with inheritance, even among FRBR Study Group members: “Expressions
inherit properties from works and manifestations inherit properties from expressions, not vice versa” (O’Neill et al., “Final Report on the Working Group on
125
©Karen Coyle CC-BY
126 / CHAPTER NINE
Aggregates”). In fact, it might be best if they did, but they don’t, as Renear and
Choi demonstrated in their paper “Modeling Our Understanding, Understanding
Our Models” (2006). In fact, the E-R modeling technique used in FRBR cannot
express inheritance. The four entities of Group 1 are linked through relationships
in a kind of bibliographic daisy chain; therefore, for example, a FRBR manifestation has access to the attributes of a FRBR work through the link to the FRBR
expression. But as described, a manifestation cannot “have” an author or subjects
because those are only attributes of a work.
In addition, the inclusion of many-to-many relationships between expression
and manifestation means that the structure of Group 1 in data modeling is by
definition a network structure. However, FRBR treats the structure as “a continuous chain,” as described in section 5.2.1:
It should be noted that although the relationships between work, expression,
manifestation, and item are depicted in the entity-relationship diagram in a
segmented way, they operate logically as a continuous chain. That is to say
that the relationship from work to expression carries through to the relationship
from expression to manifestation, and those two relationships subsequently carry
through to the relationship from manifestation to item. Thus when a relationship
is made between an expression and a manifestation that embodies the expression,
the manifestation is at the same time logically linked to the work that is realized
through the expression, given that the expression has been linked to the work it
realizes. (FRBR Final Report, 58–59)
It’s easy to understand why people assume that there is inheritance. The statement
that relationships “carry through” does sound like an inheritance model. Ignoring
the many-to-many relationships (which are illustrated in the FRBR diagrams only
through the use of multiple arrow-heads, and are easy to overlook in that form),
you can produce an instance of FRBR that looks like that shown in figure 9.1.
FIGURE 9.1
A hierarchical model
©Karen Coyle CC-BY
SOME ISSUES THAT ARISE IN FRBR / 127
This is visually the same as a taxonomy. In a taxonomy, the lower levels do
inherit from the upper levels, because each lower level is a “type of” the defined
upper level. For example, a dog is a type of mammal, so dog inherits qualities
from the super-class mammal. However, an expression is not a type of work,
and a manifestation is not a type of expression. The structure of FRBR does not
represent a taxonomy and the entities are not in a class/sub-class relationship
to each other.
The shape of the diagram is deceptive. The same diagram can, for example,
represent an organization chart, as in figure 9.2. An organization chart does
not have inheritance—lower-level employees are not “types of” the levels above
them, and do not inherit the tasks or salary of their superiors. Each employee is a
separate entity with a defined relationship to the employees “above” and “below.”
Like the FRBR entities, these relationships are not of the “type of x” or “is an
x” nature; an office worker is not a type of manager even though linked to the
manager in the organization chart.
FIGURE 9.2
An organization chart
President
Vice President
Sales
Sales Rep
Sales Rep
Vice President
Products
Manager
Human Resoures
Chief
HR Staff
Legal Staff
The four levels of Group 1 are not hierarchical in the common sense of that
word. They do have a defined order, and the order goes from concrete (item)
to increasingly abstract (manifestation, expression, work). But the entities as
diagramed are independent of each other, in the same way that a person entity
is independent of a work entity, with which it can have certain relationships.
In the formal modeling of FRBR as a Semantic Web vocabulary, the Group
1 entities are defined as “disjoint.” What disjoint means in that technology is
that the two entities cannot share any attributes or relationships. It is for this
©Karen Coyle CC-BY
128 / CHAPTER NINE
reason that each FRBR entity must have its own distinct set of attributes, like
“work title” for the work entity and “title proper” for the manifestation entity,
even though in some sense “title is a title is a title.” This disjointness also means,
though, that the entities cannot inherit attributes from each other because an
expression or manifestation is not allowed to be described with the attributes of
the work, and vice versa.
More importantly, Group 1 is not a tree structure like a taxonomy or an
organization chart. There are many-to-many relationships between the Group 1
entities, which means that Group 1 is a network of relationships. A more accurate
diagram would look like figure 9.3.
FIGURE 9.3
FRBR Group 1 with many-to-many illustrated
W
E
W
E
M
E
M
E
M
M
M
Note that the diagram in figure 9.3 only includes the primary relationships
of Group 1. You might say that they are the glue that holds Group 1 together;
they are structural in nature. The bibliographic relationships, like “adaptation
of,” “translation of,” or “supplements,” are also important, and presumably are
part of the fulfillment of user needs in the library catalog. Adding only a few of
these relationships provides a much more complex and graph-like image.
In data design, a network structure is quite common, and the many-to-many
relationship is hardly unusual. It does, however, mean that the implementation
of the conceptual model as a logical model must do some extra work to make
the many-to-many relationships function as desired.
©Karen Coyle CC-BY
SOME ISSUES THAT ARISE IN FRBR / 129
GROUP 1
ENTITIES AS “DISJOINT”
As mentioned above, the Group 1 entities defined in FRBR do not share any
attributes. The attributes are each exclusively related to a single entity, and in the
case of attributes that could be used for more than one FRBR entity, the attribute
has been defined to describe only one entity, as in work identifier, expression
identifier, manifestation identifier, or item identifier. Thus, each entity is a walled
garden, distinct from the other Group 1 entities.
Yet this separation of entities is contradicted in the FRBR Final Report’s textual description of the entities. In the text of the document, an expression is a
work as expressed, and therefore includes the content and meaning of the work.
A manifestation is a real world realization of the expressed work, and therefore
includes the content and meaning of the work as well as the expression of the
work. This alone disproves the theory that the entities are entirely separate from
each other. However, as we’ve seen, the E-R diagram and the text of the FRBR
Final Report are not perfectly aligned. The E-R diagram shows a bibliographic
description that has been broken into four dependent parts that must be reassembled to create a whole. One way to interpret this is that the text describes
the concepts, but the E-R diagrams represent a further analysis that pertains to
FRBR as a data model. However, this does not resolve the differences between
those two views.
One of the first arguments against disjointness of the FRBR entities as a
choice for bibliographic data is that, as the FRBR Final Report itself allows, not
all bibliographic communities are expected to have the same precise definitions
of the FRBR entities:
Because the notion of a work is abstract, it is difficult to define precise boundaries for the entity. The concept of what constitutes a work and where the line
of demarcation lies between one work and another may in fact be viewed differently from one culture to another. Consequently the bibliographic conventions
established by various cultures or national groups may differ in terms of the
criteria they use for determining the boundaries between one work and another.
(FRBR Final Report, 16)
Clearly, if the boundaries can be different, then the assignment of attributes
to entities must be able to differ. However, nothing in the description of the
attributes of each entity in FRBR takes this into account. Somehow, moving
©Karen Coyle CC-BY
130 / CHAPTER NINE
from page 16 of the FRBR Final Report, which describes the work, to page 32,
where the attributes are defined, all notion of possible variation is abandoned.
As Robert Maxwell says in FRBR: A Guide for the Perplexed: “Given that FRBR
emphasizes the fluidity of the concept behind the entity work, it is somewhat
surprising that the document immediately gets down to the business of defining
exactly where that line or boundary is (FRBR 3.2.1, 16–17)” (Maxwell 2008).
The FRBR Final Report’s description of the work and its list of attributes as a
single, unvarying view means that any deviation from that view is a deviation from
the FRBR model itself. The declaration of disjointness between entities enforces
the view that there is only one “right way” to model the bibliographic universe.
One must be cautious in defining entities as disjoint because disjointness
results in considerable rigidity. That rigidity makes it difficult to change (“women
cannot be priests” or “two men cannot marry”), and it also makes it difficult
for data from different communities that may not have the same restrictions to
interoperate. Disjointness means that all users of FRBR must have the exact same
definitions of the entities of Group 1 (which is most likely the purpose of the
strict definition). If a specialist community determines that a physical attribute
like color defines a new expression instead of being an attribute of manifestation,
as defined in FRBR, they will be violating the FRBR model. This rigidity, as well
as the fact that FRBR is considered “true” in its current form, means that any
application must be either FRBR or not-FRBR, thus splitting the bibliographic
world into noncompatible factions.
AGGREGATES
FRBR depicts Group 1 as a network in which each entity is in a specific relationship with one other of the Group 1 entities. They therefore form a continuous
line from work to item, and item back to work. Difficulties arise, however, in the
relationship between expression and manifestation. What often interferes here
is the complication that publishers and producers of creative works add to the
picture. Although it may be quite accurate to say that an expression is manifested
in a physical product, it is something else to say that the physical product is solely
the manifestation of the expression. The reason is that the physical, publisher-produced package nearly always has content and qualities that are in addition to the
expression. From the design of the package to liner notes, creator biographies
and prefatory material, the expression is packaged as a manifestation with content
provided by the publisher or producer.
©Karen Coyle CC-BY
SOME ISSUES THAT ARISE IN FRBR / 131
This additional content is described in detail in Gerard Genette’s Paratexts:
Thresholds of Interpretation (1997). Paratexts are all of the contents of a publication
that are not the primary text. Some are provided by the creator of the primary
content, but others are added by the product creator, the publishing house.
Paratexts are sometimes recorded in library catalog data, depending on their
perceived importance (e.g., a preface by a well-known scholar), but not always. In
cataloging, one of the more important paratexts supplied by the publisher is the
title page, which is considered the primary source of information for the catalog
record. Yet the publisher is recorded only as an element of that page, and is not
credited with any specific creative role for included content except in some rare
books or archival publications.
Not only do publishers include paratexts in publications, they also create
aggregations of expressions of works, and not always with the participation of
the original creator. Aggregates can contain works of a single author or those of
multiple authors. Because of this, the difference between multiple expressions
in a single manifestation and the presence of an expression along with paratexts
is not nearly as clear as it could be. A version of Dante’s Inferno, with a translation into English, a lengthy introduction, and copious explanatory footnotes
is hardly a mere manifestation of the expression of that work, yet there is no
physical separation between the works of Dante and his commenter that could
easily lead to the definition of two separate works. Some collections of seemingly
separate works may in fact have the intellectual characteristics of a single work,
as in publications of a professor’s lecture series. On the other hand, a recording
of two or three musical pieces by a single composer may be little more than the
publisher’s determination of an optimum number of minutes given the physical
medium of publication. A publication of separately authored papers may in some
cases cover a wide range of topics, making “workness” nearly impossible to define
for the whole. But in other cases the separate essays may have a coherence that is
hardly different from a group of chapters written by a single author.
The question that one asks at this point is whether the FRBR manifestation
refers to the expression that has been published, or whether it refers to the entire
published package. The FRBR Final Report allows for either the treatment of
an aggregate as a work in itself, or as a whole/part relationship, such as with
individual chapters in a book or articles in a journal. The whole/part relationship
has two types: the dependent part, which does not stand alone or is integral with
the whole (such as illustrations that accompany and support a text) or the independent part, such as an article in a journal. The FRBR Final Report concludes
©Karen Coyle CC-BY
132 / CHAPTER NINE
that although “there often will be no reason for a dependent part of a work to
be separately identified or described . . . independent parts of a work are more
apt to be identified and described in their own right.” (FRBR Final Report, 68)
If a FRBR manifestation does not refer to the entire publication, then it isn’t
clear how the fullness of the publication is covered in that model. It also isn’t
clear what the difference is between “is manifested as” and “is part of.” FRBR
includes part/whole relationships in its model. This, however, is where things
get complex.
Because the bibliographic description is separated into the four Group 1 entities, WEMI, it is implicit that all creative output has this structure. Thus, an
introductory essay in a published book also should have the four-part nature of
work, expression, manifestation, and item. One could take that further and consider each paratext and each design element of the finished publication to have
“WEMI-ness.” A book cover design surely would be considered a manifested
expression of a work to the extent that any other visual resource is, and the same
could be said for photographs or illustrations within the book.
In this way, the layered bibliographic description that is WEMI complicates the
part/whole relationship, because one has to decide where along the cascade of
abstractions of WEMI the part/whole relationship exists. Is this part of a work?
Part of an expression? Part of a manifestation? These are not easy questions, as
the FRBR Study Group discovered.
The FRBR Study Group created a subgroup to study the complications brought
on by aggregate publications. The Working Group on Aggregates narrowed down
the potential solutions to aggregations to two competing views: aggregates as
manifestations, which was authored by Ed O’Neil and Maja Žumer, and aggregates as works, by Barbara Tillett.
The argument by O’Neill and Žumer (undated) on aggregates as manifestations
makes use of the allowed many-to-many relationship between expressions and
manifestations in FRBR. Because any manifestation can manifest multiple expressions, an aggregate is a “combination of expressions” in a single manifestation.
They also note that there are different types of aggregates: collections (which are
aggregates of independent works), augmentations (which contain supplemental
material around a primary work or works, like the aforementioned translation and
commentary on Dante), and parallel aggregates, which are often texts that appear
in more than one language in the same publication. An aggregate as manifestation
has separate expressions of works that are manifested in a single manifestation.
Tillett’s paper (undated) takes the view that an aggregate is a work in itself,
and that it has a whole/part relationship with any works that are included in the
©Karen Coyle CC-BY
SOME ISSUES THAT ARISE IN FRBR / 133
aggregated work. Therefore, a “work can be a work of works,” which could be
shown with a recursion symbol on the box for the work entity. She states that
“The recursive symbol was omitted from the final text, but because this is causing
confusion, I feel it should be reintroduced.”
The final report of the Working Group on Aggregates (O’Neill et al. 2011)
concludes with an aggregation of its own, a combination of features from the
two viewpoints described above. The group concluded that the FRBR approach
to aggregates would use the many-to-many relationship between expressions
and manifestations. However, because aggregates have some properties of their
own (such as a creator of the aggregate), aggregates may also be seen as separate
works. Any one instance can have multiple works as well as a single aggregating
work, as shown in figure 9.4.
FIGURE 9.4
The solution to FRBR aggregates
Aggregating
Work
The Works
is realized through
is realized through
Embodied
Expressions
is embodied in
Aggregating
Expression
Aggregating
Manifestation
is embodied in
The recursive element suggested by Tillett was not included in the FRBR
Final Report. The Final Report also did not resolve the difference between a
manifestation that manifests multiple expressions and a manifestation with parts.
This seems to be a significant weakness in FRBR that needs resolution but may
require more experience with the model.
Heidrun Wiesenmüller (2011) of Stuttgart Media University studied the FRBR
aggregate model and questioned the logic of the sub-group’s conclusion. In her
view, the treatment of both individual works and aggregating works as FRBR
works at the same level did not accurately reflect the nature of those works. Her
suggestion was that there should be two levels of work, one for the aggregate
and one for the individual works. This, however, produces an even more complex
©Karen Coyle CC-BY
134 / CHAPTER NINE
set of relationships, at least in terms of how they are diagrammed. However, no
model that includes aggregations will be simple.
One solution not considered by the Study Group on Aggregates would be to
consider that all manifestations are, essentially, aggregates, comprising at least a
primary expression of a work and the paratexts accompany that content. Taking
this view would lead to a consistent treat of all manifestations since a publication
always goes beyond the mere manifestation of an expression. Even the choice of
page design, the use of covers, employing a font and pagination, perhaps adding
a table of contents and index—all of these make the manifestation more than
a manifestation of an expression. The alignment of FRBR with the CIDOC
resource model, FRBRoo, emphasizes the publication as an event in the chain
from creator to audience, and therefore injects the act of publication as external
to the creative workflow. However, FRBRoo very definitely defines aggregates as
works and does not have an equivalent to FRBR’s aggregation as manifestation.
In the view of CIDOC’s resource model, the FRBR manifestation is a compound
of content and carrier that needs to be broken apart. The work on FRBRoo may
be the best analysis of why the FRBR Study Group was unable to find a solid
solution to the problem of aggregate publications.
Before concluding, it is important to acknowledge that decisions made for the
purposes of library cataloging are not intended to solve philosophical questions
about the nature of reality. The catalog has practical goals relating to inventorying
the library collection and serving those who seek materials in the library. For that
reason it is legitimate for library cataloging rules to make decisions that serve the
library’s need, even if one could argue that they are not somehow philosophically
sound. When the members of the Working Group on Aggregates, a subgroup of
the FRBR Study Group, were tasked with formulating a solution to the problem
of how to deal with the part/whole and aggregates issue, their focus was rightly
on solving this problem for library cataloging. Whether they did so in the best way
possible is something I do not feel qualified to judge. It is unfortunate, however,
that this problem did not lead to a reevaluation of any of the assumptions on
the nature of the FRBR entities, in particular on the relationships between the
expression and the manifestation. One possible adjustment could have been to
redefine the manifestation separate from the package or container. The manifestation of the expression would then be something contained within the publication.
This would have separated the publisher’s package from the creative work it
encompasses and made it possible to describe the two independently. Instead,
the manifestation arises solely from the expression and has no individuality of its
©Karen Coyle CC-BY
SOME ISSUES THAT ARISE IN FRBR / 135
own, leaving all of the effort of publishers to be treated as a kind of afterthought.
I doubt if publishers would find that characterization flattering.
The study of aggregates is evidence of a significant difference between the
cataloging view of library metadata and the view that a data designer might take.
It is quite awkward (and not recommended) for data designs to have exceptions
or “either/or” situations. The study of aggregates as manifestations cites statistics
from OCLC showing that aggregates are only a small portion of the bibliographic
universe. However, in data modeling, it does not matter if 2 percent or 98 percent
of your instances will exhibit the characteristic in question; the model must solve
the problem in a way that is valid for all of your data. For example, in a situation
where you can have many-to-many relationships in your data, the data must be
modeled that way even though a majority of your instances might be one-to-one.
If there is the possibility to further develop the concepts first presented in FRBR,
the treatment of aggregates is an obvious area that will need additional study.
©Karen Coyle CC-BY
©Karen Coyle CC-BY
TEN
BIBLIOGRAPHIC
DESCRIPTION
AND THE
SEMANTIC WEB
C
omputer technology evolves rapidly, with change happening faster than
most humans can follow. The technology of data management evolves
less quickly than some other computer technologies because it often
requires a rather costly conversion of data and interfaces to take advantage of new
capabilities. For this reason, few can afford to jump on a new data technology
bandwagon when it first appears. Instead, most wait until such a technology has
matured before adoption.
The technology of the Semantic Web, although new to many, has been around
for almost as long as the FRBR Final Report. The first version of RDF, the basis
for the Semantic Web, was published by the World Wide Web Consortium in
1999. Adoption of RDF has been slow, but as of 2015 there is a strong movement toward implementation of this technology and integration of data with
the basic functioning of the web. Most relevant to those of us in libraries, it is
being used or is in the planning stages for a significant number of library and
archives applications.
137
©Karen Coyle CC-BY
138 / CHAPTER TEN
There is a logical progression from entity-relation modeling, such as is used
in FRBR, and Semantic Web technology that models data as things and relationships. That there is a logical progression does not mean that these technologies
are the same. In fact, as described in chapter 3 of this book, there are significant
differences between the meaning behind entity-relation models and the Semantic
Web approach that absolutely must be understood when making the transition
from one to the other. But in general, an entity-relation model is not a bad predecessor to a Semantic Web data design, as long as suitable adjustments are made.
ENTITY-RELATION MODELING AND SEMANTIC
WEB MODELS
This approach differs from some other approaches in that it begins with an
abstract of conceptual schema of the domain or universe in question. The universe is characterized in terms of the entities in it and the relationships that hold
among them. As such, the conceptual schema is not restricted by the capabilities
of any particular database system and is independent of any particular record
definition. (Tillett 1994)
To understand the Semantic Web model, we need to review how data was managed
before the Semantic Web, and how this relates to efforts to define bibliographic
data for the Semantic Web that uses FRBR as its conceptual basis. As we’ve seen,
FRBR was designed around an entity-relation analysis. The entity-relation method
was developed to support relational database management technology, which
was revolutionizing data management in the 1980s and 1990s. The problems
addressed by entity-relation design are primarily related to the efficiency of storage
and retrieval of data held in database management systems. An entity-relation
analysis normalizes data gathered for business functions with a goal of storing
each data point only once, and allowing the combination of atomized data elements to support a wide variety of business functions. At the conceptual level,
entity-relation analysis identifies the primary entities that the enterprise must
manage as data. Subsequent steps define data elements supporting or describing each entity, and which are needed to enable the tasks and workflow of the
enterprise. These analyses may support static or dynamic processes, depending
on the needs of the organization.
The Semantic Web has an entirely different approach from that used by relational databases. These differences arise from primary assumptions of the closed
world of databases versus the open world of the web.
©Karen Coyle CC-BY
BIBLIOGRAPHIC DESCRIPTION AND THE SEMANTIC WEB / 139
Databases reside generally within a system that is contained within the local
network of an organization. Although that local network most likely connects in
some way to the Internet, the database itself is not on the web. In fact, in some
cases, like banks, hospitals, and government agencies, the security of the database
is one of the key requirements of the data management function. For example,
the banking industry has its own secure network for the exchange of information
between banks; this information is not available over the open web and cannot
be directly accessed by other than local users with particular privileges.
Even for those for whom the privacy and security of the data are not paramount,
databases do not communicate directly with the web; the interaction with the
web is managed through applications that have controlled and secure access to
the database contents. This is true for library systems and their online catalogs.
Although one can access the online catalog via a web browser, the data resides
in a database on a private server.
In a Relational Database Management System (RDBMS), the data elements
allowed are controlled by the database structure definition. This structure is made
up of tables that consist of rows of data. Each data element in a table is defined
as a specific data type, such as text, integer, or structured data. The data stored
in the database is not in the same form as it appears in input or output actions.
Instead, data in the tables is combined to present views of the data for input and
output. For example, one common type of data in a database is that of names and
addresses of customers, users, or employees. The data that makes up a complete
mailing address may be stored across several tables such that city and state names
are not repeated in the database but can be linked to the particular address when
requested. The database design does not represent a record, but is a store of data
elements that are input and output in different combinations, each of which may
be considered a record by a particular user. A database may serve a wide variety
of organizational functions, including some that do not share any data elements
among them. In a large organization, management of the database is under the
control of a technical department, although input and output may take place
throughout the organization by different types of employees. These employees
may be entirely unaware of the nature of the whole database, and only see the
portion that is relevant to them through a user interface.
A common aspect of relational databases is that they have features built in for
data quality control. In fact, the database design itself forces data to conform to
rules. For example, certain data elements can be required in the database design.
If the database has only one storage point for a data element, more than one
element simply cannot be added. In particular, no new element types can be
added by database users until changes are made to the database design. Because
©Karen Coyle CC-BY
140 / CHAPTER TEN
these changes often require a significant amount of planning and testing, data
stored in databases has a tendency to change only with a certain effort. Anything
not included in the RDBMS design is not available to the applications that will
use the data.
It is common to hear the web of linked data referred to as a way to use the
web as your database. Although this analogy holds, the difference between the
web as database and the closed, controlled world of the enterprise database
management system is enormous. The open world of the Semantic Web has no
predefined boundaries. The maxim used on the Semantic Web is that “anyone
can say anything about anything.” The Semantic Web has no concept of quality
control analogous to an RDBMS. This has advantages as well as some challenges.
The advantage is that new data types can be added as needed. Rather than the
slow and often painful process of database modification that traditional closed
databases require, the web of linked data can allow new concepts to be added by
anyone at any time. Any data on the web can be linked to your data, just as today
anyone on the web can add a document that links to a web page of yours. There
is also no way to require that certain data always be present in the Semantic Web
model. Because the open world of the web is considered to be ever-changing (as
is the real world of information), it is not possible to predict what data will be
available at any given moment in time. Data that is missing at one moment may
be present a moment later. Therefore the Open World Assumption, one of the
fundamental tenets of the Semantic Web, does not treat missing data as an error
but merely as something momentarily unknown.
In the pre-Semantic Web data environment, we generally consider ourselves
to have data encapsulated in records outside of the database. The database itself
has tables and does not conform to the form of the records that are input and
output. However, for those who create and use the data, the record is the unit that
defines an appropriate grouping of data elements. Like a database, the Semantic
Web does not have records. But the input and output, when using the Semantic
Web standards based on RDF, also are not packaged as records. Regardless of
the view that one has at any given time, the Semantic Web is always composed
of three-part statements called triples, which are autonomous atomic statements
that can be combined in any desired configuration. A search against the Semantic
Web returns some number of triples that match the search query.
FRBR, and subsequently RDA, were modeled as one would model data for
an RDBMS. As part of the development of RDA, the Joint Steering Committee
for RDA produced a document that presented three high-level database models
for bibliographic data (Delsey 2009). The most advanced of these models, and
©Karen Coyle CC-BY
BIBLIOGRAPHIC DESCRIPTION AND THE SEMANTIC WEB / 141
presumably the preferred one, was called “Scenario 1,” and made use of concepts
from relational database design. The documents reflect some assumptions about
systems efficiency and user service capabilities that were held by members of the
committee:
The data structures used to store the data and to reflect relationships, however,
will have a bearing both on the efficiency of data creation and maintenance, and
on the ease and effectiveness with which users are able to access the data and
navigate the database. (Delsey 2009)
There are two ways in which efficiency of a database is commonly measured: the
time needed for operations to the database to complete, and the space needs
to data storage. In practice these two considerations are often in conflict. It is
assumed that the storage requirements for FRBR work data (and to a lesser extent
FRBR expression, because it contains fewer fields) would be reduced because
there would be less duplication of these elements in the database. None of the
statistics that we have on hand, however, measure data storage, so it remains an
open question whether there are significant savings.
In terms of time necessary to complete operations, the efficiency of a database
often depends on how many joins and reads are required. Joins and reads are
what bring together related data that is stored in the separate database tables.
Some efficiency that is gained by reducing redundancy is lost when multiple
tables must be included in a single database activity. Database models are adjusted
to meet these efficiency requirements by testing. This type of testing would
be needed before any definite statements could be made about a FRBR-based
database model.
Suffice it to say that at this time there are no studies available that support any
claims of efficiency for RDA as a FRBR-based bibliographic model. How efficient
the sharing of works and expressions for catalog maintenance depends on the
extent of redundancy of entities in the local catalog, which is usually a function
of a library’s mission and size, as well as the extent to which bibliographic data
is held in a shared data pool rather than copied to the local library system.
The Semantic Web model is also seen as holding promise in terms of data
sharing. The Semantic Web is by its nature a shared “cloud,” although the same
technologies are used in closed enterprise systems. The primary data structure
of the Semantic Web, RDF, is already being used in large corporations that need
to share and link data between operations and over large geographical spaces.
There is tangible tension between the Semantic Web design for the open web
©Karen Coyle CC-BY
142 / CHAPTER TEN
and the use of these technologies in closely held data environments. For many
enterprises, it’s not a matter of entirely open or entirely closed but, as with traditional databases, using the networking capabilities of the Internet to share select
information with a wider public.
FRBR IN RDF
The conceptual model of FRBR and variants on that model have been expressed
in recent RDF vocabularies developed for library data. FRBR has also been used
in bibliographic models designed outside of the library. There are also Semantic
Web implementations of bibliographic data, such as some develop for academic
citations, that do not make use of the FRBR conceptual entities. This section
focuses on some of the projects that have transformed the entity-relation model
of FRBR to a Semantic Web vocabulary.
FRBRer
In 2011, the IFLA FRBR Review Group, the group that now maintains the
FRBR standards (including functional requirements for authority and subject
data) issued its official version of FRBR in RDF. Called FRBRer, with the “er”
standing for entity-relation, this version is based on the 2009 edition of the FRBR
Study Group’s final report.
FRBRer is an encoding of FRBR as defined in the document. FRBRer uses
the World Wide Web Consortium’s standard for the definition of vocabularies,
the Web Ontology Language (OWL), which is described in chapter 3 of this
volume. As the literal translation of an entity-relation model, FRBRer does not
follow some practices that are common in Semantic Web vocabularies. In particular, FRBRer does not make use of super- or sub-classes to define logical types
of entities. Without class relationships, which are heavily used in Semantic Web
vocabularies, FRBRer cannot define a single relationship that is valid between
a work and any members of FRBR Group 2. Instead, FRBRer must develop a
specific relationship for each individual Group 2 entity: for example, “is created
by person,” or “is created by corporate body.” This must be done for each of
the relationships involving Group 2 entities such as “realized by,” “produced
by,” and “owned by.” The later addition of “family” to Group 2 meant that
new relationships had to be added for it as well. This has an effect on the ease
of extensibility of FRBRer: with class relationships it would only be necessary to
define the entity “family” as a sub-class of the class “responsible entity,” and family
©Karen Coyle CC-BY
BIBLIOGRAPHIC DESCRIPTION AND THE SEMANTIC WEB / 143
would be immediately usable with all relationships defined between Group 2 and
Group 1. As it is, the addition of family also requires the family-specific definition
of each of the relevant relationships to the other entities.
It is clear that FRBRer is designed with a closed-world point of view rather than
the open view of the Semantic Web. This view in inherited from the entity-relation origins of FRBR, because entity-relation modeling is a design method for
database technologies and not for the Semantic Web. As I noted in the chapter on
technology, there is a common misinterpretation of the Semantic Web vocabulary
definition language OWL that reads the OWL rules as quality-control constraints
on the data rather than as axioms for making inferences from the data. FRBRer
is based on this misinterpretation.
One way that FRBRer affirms its closed-world view is by declaring all entities
and all of the FRBR attributes as disjoint from each other. This means that it isn’t
possible for anyone using FRBRer to create data that varies in the assignment of
attributes to entities. The result of this is that the data coded in this vocabulary
is very fragile, with any deviation from the defined terms causing OWL-aware
software to fail to function as desired. This fragility is not easy to mitigate because
OWL does not have any functions that enforce quality control on data; instead,
in the open world where “anyone can say anything about anything,” vocabularies
need to be as forgiving as possible. The strictness implied by disjoint classes may
be unrealistic in an imperfect world, with a few exceptions for undeniable truths,
such as “up” being disjoint “down.” Nearly any category has exceptions that
need to be handled, even the fundamental states “solid, liquid, or gas,” which
include intermediate transitions that can be more than one state at the same time.
In the open world of the web, the strictness implied by disjoint classes generally
prevents this data from interoperating with any data that is based on a different
model, even if they have linkable elements in common. In the case of FRBRer,
this creates an incompatibility with any bibliographic data that does not precisely
separate its bibliographic description into identically defined entities. Not only
does this mean that variants of FRBR as desired by some library specialist and
non-book communities cannot be used alongside conformant FRBRer, but that
FRBRer on the open web may not be able to link to similar data using models
like FRBRcore, BIBFRAME, or even RDA in RDF.
In this way the primary advantage of the Semantic Web, discovery across
heterogeneous data contributed to the web of data by different communities,
is negated by the definition of vocabularies using OWL with an inappropriate
closed-world assumption. FRBRer, as defined, is a vocabulary that will necessarily
be usable only in its own silo.
©Karen Coyle CC-BY
144 / CHAPTER TEN
RDA in RDF
Resource Description and Access (RDA), the cataloging rules adopted in 2013
by many North American libraries, also has a defined RDF-based vocabulary. The
development of RDA was closely tied to the bibliographic concepts presented
in the FRBR Final Report, and the structure of the RDA rules follows that of
the FRBR groups—primarily Group 1, which is the main focus of descriptive
cataloging. A list of data elements, each with their assigned FRBR entity, was
published as a supplement to the final text of RDA. After a 2007 meeting between
volunteers active in the Dublin Core Metadata Community and members of
the RDA development group, the RDA elements were defined in RDF on an
experimental basis. This activity preceded the development of FRBRer, and RDA
in RDF was based on a very loose vocabulary of FRBR entities as RDF classes.
Although FRBRer is now available, and there is an official version of RDA in
RDF that is managed by the Joint Steering Committee (JSC) that maintains
the cataloging rules, the JSC has so far chosen not to follow the vocabulary as
defined by FRBRer. Instead, RDA in RDF has created its own FRBR classes to
define the domain of each of the data elements, but does not attempt to impose
E-R-like rules on the resulting bibliographic description. Each RDA element is
also sub-classed to a parent element that is not associated with any FRBR entity,
creating a FRBR-neutral version of the vocabulary that may be more acceptable to nonlibrary communities for whom the Group 1 entities of FRBR are
unfamiliar, and perhaps even not useful. RDA in RDF creates a super-class for
person, corporate body, and family that is called “agent,” but no super-class for
the Group 1 entities. Recall that the FRBR Review Group has made clear that in
their analysis the groups are not to be represented as super-classes; instead, they
exist as organizing elements in the documentation only.
With an awareness of the open world and the fact that there are many sources
of bibliographic data, both within the library community and outside of it, RDA
in RDF proposes linking RDA elements to commonly used terms from other
vocabularies using sub-class relationships, with a more general vocabulary like
Dublin Core Metadata Terms as the common language. This can bridge the gap
between the over 900 elements in RDA and bibliographic data as approached by
nonlibrarians. There probably is no other community that has so many different
types of titles (key title, parallel title, series title, parallel series title, etc.), yet most
communities creating bibliographic data will have an element that is compatible
with Dublin Core “title,” or that can be sub-classed to it. Because the primary
goal of the Semantic Web is to allow linking between comparable data across the
web, it is a good idea for everyone to design links into their community-specific
©Karen Coyle CC-BY
BIBLIOGRAPHIC DESCRIPTION AND THE SEMANTIC WEB / 145
vocabulary to well-known vocabularies used on the web. In this way one avoids
being stuck in a silo where library data can only connect to other library data
on the web.
That said, RDA in RDF itself has few relationships between terms, and few uses
of classes and sub-classes. This is possibly a reflection of its origins as a list of terms
derived from the thirty-odd chapters of the RDA cataloging rules. You could say
that RDA has not yet been subjected to the data design phase that would look
at the desired functionality that could be delivered with complex bibliographic
data expressed in RDF. This type of design often begins with a statement of use
cases: what is it that we want to do, and that we could do, with this data? Recall
that in chapte 2 on modeling we had simple use cases like “find a book of which
the author is known.” Given the kinds of capabilities that we have with current
data management technology, including the possibility of using the entire web
as the context for our information services, the number and kinds of use cases
would surely grow. Do we want to promote linking from online sites for readers,
like GoodReads or LibraryThing, to library materials? From the music database
MusicBrainz to library collections of recorded sound? From the Internet Movie
Database to film descriptions in library catalogs? Do we want to incorporate more
information about authors alongside the library’s bibliographic holdings? Do
we want to continue to store most catalog data locally, or would “cloud-based”
sharing of data be more efficient? The list would end up being quite long, but
without an exploration of goals, we cannot make rational decisions to guide our
development.
BIBFRAME
The RDF vocabulary of BIBFRAME, which is being developed simultaneously
by Library of Congress, Zepheira, and a number of library projects, is not an
implementation of FRBR, but is clearly influenced by the FRBR model. It has
a two-entity model of bibliographic description, with the entities called work
and instance. The BIBFRAME work represents the content portion of the bibliographic description, and the instance describes the carrier. Generally speaking,
the BIBFRAME work encompasses attributes that are associated with both the
FRBR work and expression; the BIBFRAME instance is analogous to the FRBR
manifestation. Item-level information is not treated as one of the primary bibliographic entities in BIBFRAME. BIBFRAME also resembles the pre-FRBR
data model with a central bibliographic description plus authority entities that
are similar to name and subject authorities (figure 10.1).
©Karen Coyle CC-BY
146 / CHAPTER TEN
FIGURE 10.1
BIBFRAME model
subject
creator
WORK
hasinstance
INSTANCE
publisher
format
publishedAt
BIBFRAME and FRBR come out of different communities within the library
environment. Although the FRBR Final Report provides background on the
bibliographic theory that led to the creation of FRBR, the link between standard
cataloging practice and BIBFRAME is less clear. As the “named successor” to
the current standard bibliographic record, MARC 21, BIBFRAME is primarily
a data standard. To accommodate MARC 21, BIBFRAME will necessarily have
more detail than the FRBR conceptual model, which was developed ostensibly
as a minimum set of bibliographic data.
©Karen Coyle CC-BY
BIBLIOGRAPHIC DESCRIPTION AND THE SEMANTIC WEB / 147
SEPARATE WORK AND INSTANCE
WITH LINKS
SINGLE RESOURCE WITH
WORK AND INSTANCE DATA ELEMENTS
ex:ResourceA
ex:ResourceA
bf:workTitle ex:AdventuresOfTomSawyer ;
bf:creator lcna: n79021164 ;
bf:hasInstance ex:ResourceB ;
bf:workTitle ex:AdventuresOfTomSawyer> ;
bf:creator lcna: n79021164 ;
bf:language is0639–2:eng ;
bf:language is0639–2:eng .
bf:instanceTitle ex:TheAdventuresOfTomSawyer> ;
ex:ResourceB
bf:providerDate “1996” .
bf:providerDate “1996” ;
bf:instanceOf ex:ResourceA ;
bf:instanceTitle ex:TheAdventuresOfTomSawyer .
BIBFRAME’s vocabulary is more compatible with the open web and with
potential variations in bibliographic concepts than is FRBRer. BIBFRAME makes
use of classes and sub-classes in ways that are convenient for designers of systems,
which will probably facilitate searching and other system functionality. It does
not define any classes as disjoint. Because of this, both of the above examples
are valid instances of the BIBFRAME vocabulary, with no loss of information:
The above examples illustrate that the storage of descriptive elements in separate work and instance units is not required in BIBFRAME. Whether you keep
work and instance separate or not can depend on your needs. In fact, at times
data can be stored without separate works and instances, yet can be transmitted
as separate BF entities when desired because the entities can always be created
from a properly defined RDF vocabulary. This is an important lesson about RDF,
especially as compared to the data models that we are most familiar with: in RDF,
the meaning of an element is in the defined vocabulary, not in a record structure.
You can define data that adheres to the concepts of work and instance, or even
of work, expression, manifestation, and item, without that predetermining the
structure of your data. Whatever view of your data you work with at any given
moment depends only on what works best for you for that function.
ALTERNATIVE MODELS USING FRBR CONCEPTS
In previous chapters we looked at how the reality of FRBR is different from its
stated goals: both the goals that led to the creation of the FRBR Study Group
and the goal presented in the FRBR Final Report itself. Yet there is no question
that FRBR, and in particular the Group 1 entities, resonates with many people
©Karen Coyle CC-BY
148 / CHAPTER TEN
both within librarianship and outside of it. We see a general recognition that the
bibliographic resource is a complex thing that can be approached from a number
of different points of view. This, at least, seems to garner wide agreement. What
is not agreed upon, however, is a single interpretation of that complexity. It
seems that many have a need to express the complexity of what they describe
bibliographically, but those needs have a great number of potential expressions.
Those who have made use, in their own ways, of FRBR concepts have employed
them to reinterpret the bibliographic entity as a cascade of abstractions, from the
most abstract work through some number of intermediary levels until the actual
physical item is described. The number of variants of this path, however, seem
to be without limit, nor do the multi-entity models even agree on the nature of
the levels of abstraction. FRBR itself is a shallow model, with a set of entities but
no sub-entities. In some cases, those who borrow FRBR concepts extend the
model significantly beyond its flat nature. In other cases, the model is reduced
in the number of entities while at the time it is given greater depth through the
creation of a hierarchy of entities.
Because it is a conceptual model, there is nothing in FRBR that has not
been treated as open to reinterpretation, not even the seemingly uncontroversial
concept of physical item. Where one draws the line between the physical and
bibliographical is not as clear as you might think. FRBR, the library, and archivebased models often combine physical description (e.g., dimensions of the package)
with a description of what is printed on the package itself. This is a mixture of
information that harks back to the catalog record that presents a single line or
field that includes extent, illustrative matter, and size (“xiii, 368 p. : ill. ; 24 cm.”)
Arguably a more logical separation would focus solely on physical properties.
However, defining what is relevantly physical regarding digital resources, is not
easy, and is even more difficult in a mixture of hard copy and digital resources,.
Each of the models presented below is a response to FRBR with some alterations.
Given that FRBR presents itself as a conceptual model (at least in the text of the
document), each of these uses the conceptual model and proposes a direction or
implementation of it under different assumptions or using different technologies.
Included here is just a selection of models that have riffed on FRBR’s melody, or have
come to a similar conclusion independently. Each of these confirms some aspects
of the multi-entity bibliographic model, while drawing into question some others.
FRBRcore
The first development of the FRBR conceptual model in RDF was done in 2005
by Ian Davis and Richard Newman, in an effort that was not supported by the
©Karen Coyle CC-BY
BIBLIOGRAPHIC DESCRIPTION AND THE SEMANTIC WEB / 149
IFLA group. This version of FRBR was called “FRBRcore” because it included
the entities and relationships of FRBR but not the attributes. This may seem
odd, as the attributes seem to have the greatest importance because they carry
the descriptive information about bibliographic resources. This approach, however, makes sense if the role of classes in RDF is understood. RDF classes are
similar to entities in a conception entity-relation model, and therefore entities
are often designed as classes when models are interpreted in RDF. The classes in
FRBRcore provide a conceptual framework that can be extended as needed. As
we’ll see when we discuss alternate models based on FRBR, a single conceptual
framework can be used as the basis of some very different solutions. FRBRcore
makes those solutions possible.
Davis and Newman were familiar with library models but they also were
involved in RDF implementation. Although they were the first to develop an
RDF version of FRBR, their vocabulary was never endorsed by the IFLA Working Group on the Functional Requirements for Bibliographic Records. Indeed,
FRBRcore varied some from the E-R model in the FRBR Final Report. Davis
and Newman added a super-class that encompasses the entities of Group 1,
which they called “endeavor.” They also created super-classes for Groups 2 and
3, “responsible entity” and “subject,” respectively. In addition, they assigned a
super-class, “spatial thing,” to both object and place, because both of these could
require geo-location attributes. The class structure in RDF is used in applications
to address data at different levels of specificity. For example, the class “responsible entity” becomes shorthand for “person or corporate body” in program
functions. The total number of entities in FRBR are relatively small, but in many
RDF vocabularies the classes outnumber the descriptive elements, and are vital
for efficient processing of the data.
FRBRcore could be considered experimental in nature, and is definitely not
intended to be complete. The authors appear to have added a few elements
intended to illustrate how FRBRcore could be extended. As sub-classes of work,
they defined ClassicalWork, LegalWork, LiteraryWork, and ScholarlyWork. Each
of these is a type of the more general concept of work. This extension of the work
in FRBRcore may be surprising to those accustomed to library bibliographic data,
because nowhere in FRBR or in any of the cataloging rules is there a discussion of
types of works similar to those devised by Davis and Newman. The nearest thing
is the FRBR attribute form of work, which includes as examples “play, poem, and
novel.” FRBR form of work isn’t the same as the FRBRcore subtypes of work;
however, either or both could be the basis for extending the work concept. In
fact, a concept like work could be extended in a number of different directions,
because in RDF classes are not exclusive. There is also no limitation in RDF on
©Karen Coyle CC-BY
150 / CHAPTER TEN
the creation of a super-class that is a generalization of one or more classes. That
FRBRcore has the super-class “endeavour” does not mean that the FRBR entities
of Group 1 are no longer valid or that their meaning has changed.
FRBRcore is highly visible today in the linked data cloud. The Linked Open
Vocabularies project reports that FRBRcore is used in twelve vocabularies, and
appears in nearly 30 million instances, although most uses are from the union
catalogs of Bavaria, Berlin, and Brandenberg, and the union catalogs of Hessen
and parts of the Rhineland. These catalogs account for about 24 million instances
of FRBRcore in RDF triples.
FRBRoo’s Object-Oriented Model
FRBRoo is a harmonization of the FRBR entity-relation model and the Conceptual
Reference Model of the International Council of Museums (CIDOC CRM). It
is, as its name indicates, an object-oriented data model, but the vocabulary has
also been defined in RDF. FRBRoo is not intended as a replacement for FRBRer,
but as an interpretation for an object-oriented environment and a harmonization
with museum practice. The FRBRoo document explains that its project also serves
as a proof of concept of the FRBR model:
Expressing the FRBR model in a different formalism than the one in which it
was originally developed provides a means to evaluate the model in terms of
its internal consistency. It is also a good opportunity to correct some semantic
inconsistencies or inaccuracies in the formulation of FRBR that may be regarded
as negligible when FRBRer is only used in a library catalogue context, but that
prove to be quite crucial from the moment one strives to design an overall model
for the integration of cultural heritage related information. (Doerr 2006)
Object-oriented models define both things and processes, while the E-R model
used in FRBR is a static definition of entities and relations. However, some of
the relations in FRBR (e.g., “manifests” or “expresses”) imply some action, and
FRBRoo has taken an action or event-oriented view. FRBRoo includes an event
for each of the WEMI concepts, such as Work Conception, Expression Creation,
Carrier Production Event, and Publication Event. It recognizes the importance
of these events and the role played by the various actors that are involved in the
transformation from a creative concept to something shared with others.
©Karen Coyle CC-BY
BIBLIOGRAPHIC DESCRIPTION AND THE SEMANTIC WEB / 151
FIGURE 10.2
A segment of the FRBRoo model—From Work to Expression, dynamic view
R18 created
(was created by)
E5 Event
E28 Conceptual Object
F4 Manifestation
Singleton
R20 recorded (was recorded through)
E65 Creation
R16 initiated (was initiated by)
F27 Work Conception
F1 Work
R19 created a realization of
(was realized through)
F14 Individual Work
F28 Expression Creation
F16 Container Work
R17 created
(was created by)
F19 Publication Work
R22 realized (was
realized through)
F21 Recording Work
F29 Recording Event
F20 Performance Work
F2 Expression
R21 created
(was created by)
E7 Activity
R23 created
a realization of
(was realized through)
F30 Publication Event
F22 Self-Contained Expression
F25 Performance Plan
R25 performed
(was performed in)
F31 Performance
R24 created
(was created through)
F26 Recording
F24 Publication Expression
©Karen Coyle CC-BY
152 / CHAPTER TEN
FRBRer envisions bibliographic entities as static, ever-existing things that come
from nowhere, and overlooks the complicated path from the initial idea for a
new work in a creator’s mind to the physical item in a user’s hands through the
dramatically important decision-making on behalf of publishers. As a matter of
fact, bibliographic records do contain implicit information about that complicated
path and the relationships it implies between and among bibliographic objects;
FRBRoo digs that implicit information out of bibliographic structures, e.g. the
precise meaning of “date of publication.” (Doerr 2006)
FRBRoo also extends the FRBR entities using super- and sub-classes, as can be
seen in figure 10.2. For example, there is an abstract class super to work and
expression called “conceptual object,” and sub-classes under work itself for types
of works, such as “individual work,” “complex work,” and “publication work.”
Significantly, FRBRoo recognizes publication as an action, and treats Publication
Work as an entity sub-classed to Container Work. This at least partially responds to
the difficulty that the FRBR Review Group had with aggregates, and smooths the
transition from the expression to a publicly available package that has physical and
intellectual characteristics that are added by a publisher but that are not included in
the expression itself. In FRBRoo, every publication is a kind of aggregate, because
it always contains some aspect of creation added by the publisher. This makes a
publication, by definition, a package of multiple creation activities. Works that are
not published, such as works of art, can be manifested without the intervention
of a publisher, and therefore do directly manifest what has been expressed. In the
2006 version of the FRBRoo documentation, edited by Martin Doerr and Patrick
Le Boeuf, published works were likened to car models in their relationship between
the original creator and what comes out of a manufacturing process.
“Manifestation” can be two completely different things: Either it is an industrial
product, i.e., a Type, like a particular car model, or it is a Physical Man-Made
Thing that was produced as a unique carrier of an Expression. Industrially printed
books belong to the first category, and are indirectly related to the main author’s
original creations. (Doerr 2006)
This echoes some of my own concerns about FRBR’s treatment of the manifestation as being in a direct line from the expression of the work, without recognition
of the many aspects of the published resource (book, sound recording, or film)
that are contributed by the publisher.
©Karen Coyle CC-BY
BIBLIOGRAPHIC DESCRIPTION AND THE SEMANTIC WEB / 153
In the course of discussion however it was recognized, that virtually any book
is composed of multiple, distinct works: the text, the illustrations, the editors
work on lay-out, type phase etc. The latter was widely ignored in FRBR. . . .
This situation demanded for a general model explicating both the individual
contribution and the unity of the integrated product. (Doerr 2007)
FRBRoo appears to have a consistent grounding in object-oriented technology,
and to be shepherded by a group that understands that technology. That said,
FRBRoo is extremely complex, and understood by few. If it is to be adopted it
will need a user-facing interface that is comprehended by a wide range of metadata producers.
<indecs> Event-Oriented Model
The <indecs> metadata model was developed in the late 1990s, and the
current revised version is from 2000. <indecs> models metadata for intellectual resources around an e-commerce viewpoint, but intends its design to
interoperate with the full product flow from the original creator, through the
publishing and manufacturing steps, to outlets like stores and libraries, and
then to the end-user.
<indecs> takes an event-based view of the metadata model, and defines metadata as a “relationship that someone claims to exist between two entities.” All
relationships are events, either static or dynamic, and events have inputs, agents,
and outputs. The agents are key in the <indecs> model because these inform
the question of intellectual property rights, which are one of the main elements
of commerce.
The treatment of each entity as being the result of an event with the involvement of an agent resolves some of the questions about the nature of the FRBR
Group 1 entities: where FRBR has works, expressions, and manifestations as primary entities with flat relationships and no intervening human activity, <indecs>
includes the persons or agents that act to create the entities (figure 10.3). This
answers the question: how does a work get expressed, and by whom?
<indecs> also recognizes that different media and different products can have
different events. In particular, the performance event for music produces types
of expressions that are not common for textual works. <indecs> places no limits
on the types of relations that can be included in the model, although it lays out
a general framework of metadata properties.
©Karen Coyle CC-BY
154 / CHAPTER TEN
FIGURE 10.3
The <indecs> model of bibliographic metadata
is used in
percept
artefact
place
happens in
manifestation
produces
has instance
item
is fixed in
party
acts in
event
expression
is abstracted to
is abstracted to
is expressed in
time
abstraction
happens in
is part of
is used in
concept
FaBiO FRBR-Aligned Bibliographic Ontology
FaBiO is one of the vocabularies defined in the Semantic Publishing and Referencing (SPAR) suite. This vocabulary development project has scholarly publishing as
its main focus, with coverage primarily aimed at works that are published, textual,
and/or referred to by bibliographic references. It defines its own set of entities
that are sub-classed to the original FRBR entities (work, expression, etc.) using
the FRBRcore RDF vocabulary. FaBiO’s main approach to the FRBR entities is
to use them as classes for types of scholarly publications. FaBiO sub-classes the
FRBRcore work class with almost thirty different subtypes, including biography,
reference work, dataset, and sound record. The FRBRcore expression class yields
over fifty sub-classes, among which are chapter, editorial, presentation, spreadsheet, and Gantt chart.
In keeping with its approach of creating sub-classes of the FRBR entities
defined as classes in FRBRcore, a manifestation in FaBiO is defined as having
three sub-classes: analog manifestation, digital manifestation, and manifestation
collection. The model is illustrated in figure 10.4.
©Karen Coyle CC-BY
BIBLIOGRAPHIC DESCRIPTION AND THE SEMANTIC WEB / 155
FIGURE 10.4
FaBiO model showing expression and manifestation examples
W
type
creator
type
Research paper
M
“Yves Marcoux”
pub
date
print
“Wiley”
“09-2009”
E
type
title
date
DOI
vol.
no.
Journal article
type
“Intertextual semantics”
2009
10.1002/asi.s1134
M
format
date
digital
pdf
“21-08-2009”
9
60
Along with the classes derived from FRBR entities, FaBiO has dozens of
properties for bibliographic description, few of which would be considered exact
equivalents of descriptive elements in library data. There are some common bibliographic properties like title, publisher, data of publication, pages, identifier, and
language, but FaBiO also has an extensive list of dates relating to the workflow
of academic publishing, such as “has submission date,” “has embargo date,”
and “has date received.” FaBiO does not, however, align use of bibliographic
properties to specific sub-classes of FRBR’s WEMI. Instead, the descriptive
properties are associated with the super-class that represents the bibliographic
resource as a whole.
In addition to the extensions of FRBR provided by FRBRcore, FaBiO creates
direct links between works and manifestation, works and items, and expressions
and items.
Where FaBiO most notably diverges from a library cataloging interpretation
of the entities, though, is in its emphasis on the expression. Like the expression-dominant model of Shoichi Taniguchi, FaBiO interprets the manifestation
as a physical carrier, with all of the content description properties being associated
with the expression. This model defines different digital formats of the same
publication or manuscript in the item entity, which is a possible resolution to the
©Karen Coyle CC-BY
156 / CHAPTER TEN
situation that libraries face with electronic books and other electronic resources
where only the digital format varies.
Summary
FRBRer, FRBRcore, FaBiO, <indecs>, BIBFRAME, and RDA in RDF are variations on a theme. They have much in common, but each has its own approach.
webFRBRer defines a strict closed-world interpretation of the FRBR model. RDA
in RDF implements the elements of the FRBR-based cataloging rules. FRBRoo is
highly sophisticated in its design but is so far unconnected to mainstream library
cataloging. FRBRcore and BIBFRAME show promise as vocabularies for the open
web, and FRBRcore has been the basis for the development of other bibliographic
vocabularies. BIBFRAME, however, is taking on the unenviable task of carrying
forward into RDF the centuries-old practices of traditional library cataloging
and in particular that tradition as a coded machine-readable record, MARC 21.
None of these can be considered mature at this time, and the future
of bibliographic data in RDF is still at an experimental stage. Although there
is some speculation in the library community about which of these models will
prevail, we have to consider the possibility that there will be more than one
model in use. Already there are efforts to assure that bibliographic models under
development promote interoperability, not only among the library, archives, and
museum bibliographic models, but also with the worlds of publishing, academe,
and the reading public.
©Karen Coyle CC-BY
AFTERWORD
T
here is no question that FRBR represents a great leap forward in the
theory of bibliographic description. It addresses the “work question” that
so troubled some of the great minds of library cataloging in the twentieth
century. It provides a view of the “bibliographic family” through its recognition
of the importance of the relationships that exist between created cultural objects.
It has already resulted in vocabularies that make it possible to discuss the complex
nature of the resources that libraries and archives gather and manage.
As a conceptual model, FRBR has informed a new era of library cataloging rules.
It has been integrated into the cataloging workflow to a certain extent. FRBR has
also inspired some nonlibrary efforts, and those have given us interesting insight
into the potential of the conceptual model to support a variety of different needs.
The FRBR model, with its emphasis on bibliographic relationships, has the
potential to restore context that was once managed through alphabetical collocation to the catalog. In fact, the use of a Semantic Web technology with a
model of entities and relations could be a substantial improvement in this area,
because the context that brings bibliographic units together can be made explicit:
“translation of,” “film adaptation of,” “commentary on.” This, of course, could
be achieved with or without FRBR, but because the conceptual model articulates
the relationships, and the relationships are included in the recent cataloging rules,
it makes sense to begin with FRBR and evolve from there.
However, the gap between the goals developed at the Stockholm meeting in
1991 and the result of the FRBR Study Group’s analysis is striking. FRBR defined
only a small set of functional requirements, at a very broad level: find, identify,
select, and obtain. The study would have been more convincing as a functional
analysis if those four tasks had been further analyzed and had been the focus of
the primary content of the study report. Instead, from my reading of the FRBR
Final Report, it appears that the entity-relation analysis of bibliographic data took
precedence over user tasks in the work of the FRBR Study Group.
157
©Karen Coyle CC-BY
158 / AFTERWORD
The report’s emphasis on the entity-relation model, and the inclusion of three
simple diagrams in the report, is mostly likely the reason for the widespread belief
that the FRBR Final Report defines a technology standard for bibliographic
data. Although technology solutions can and have been developed around the
FRBR conceptual model, no technology solution is presented in the FRBR Final
Report. Even more importantly, there is nothing in the FRBR Final Report to
suggest that there is one, and only one, technology possible based on the FRBR
concepts. This is borne out by the examples we have of FRBR-based data models,
each of which interprets the FRBR concepts to serve their particular set of needs.
The strength of FRBR as a conceptual model is that it can support a variety of
interpretations. FRBR can be a useful model for future developments, but it is a
starting point, not a finalized product.
There is, of course, a need for technology standards that can be used to
convey information about bibliographic resources. I say “standards” in the plural,
because it is undeniable that the characteristics of libraries and their users have
such a wide range of functions and needs that no one solution could possibly
serve all. Well-designed standards create a minimum level of compliance that
allows interoperability while permitting necessary variation to take place. A good
example of this is the light bulb: with a defined standard base for the light bulb
we have been able to move from incandescent to fluorescent and now to LED
bulbs, all the time keeping our same lighting fixtures. We must do the same for
bibliographic data so that we can address the need for variation in the different
approaches between books and non-books, and between the requirements of the
library catalog versus the use of bibliographic data in a commercial model or in
a publication workflow.
Standardization on a single over-arching bibliographic model is not a reasonable solution. Instead, we should ask: “what are the minimum necessary points
of compliance that will make interoperability possible between these various
uses and users?” Interoperability needs to take place around the information and
meaning carried in the bibliographic description, not in the structure that carries
the data. What must be allowed to vary in our case is the technology that carries
that message, because it is the rapid rate of technology change that we must be
able to adjust to in the least disruptive way possible. The value of a strong conceptual model is that it is not dependent on any single technology.
It is now nearly twenty years since the Final Report of the FRBR Study Group
was published. The FRBR concept has been expanded to include related standards for subjects and for persons, corporate bodies, and families. There is an
ongoing Working Group for Functional Requirements for Bibliographic Records
©Karen Coyle CC-BY
AFTERWORD / 159
that is part of the Cataloguing Section of the International Federation of Library
Associations. It is taken for granted by many that future library systems will carry
data organized around the FRBR groups of entities. I hope that the analysis that
I have provided here encourages critical thinking about some of our assumptions,
and fosters the kind of dialog that is needed for us to move fruitfully from broad
concepts to an integrative approach for bibliographic data.
©Karen Coyle CC-BY
©Karen Coyle CC-BY
BIBLIOGRAPHY
ALA Committee on Co-Operation. “Standard Catalogue Card.” Library Journal 1,
no. 8 (1877): 285.
Albertsen, Ketil, and Carol van Nuys. “Paradigma: FRBR and Digital Documents.”
Cataloging and Classification Quarterly 39, no. 3–4 (2005): 37–41.
doi:10.1300/J104v39n03_09.
Andersen, Jack. “Materiality of Works: The Bibliographic Record as Text.” Cataloging and Classification Quarterly 33, no. 3/4 (November 2013): 39–65.
doi:10.1300/J104v33n03_04.
Antelman, Kristin. “Identifying the Serial Work as a Bibliographic Entity” Library
Resources and Technical Services 48, no. 4 (October 2004): 238–55.
Armstrong, Sharon Lee, Lila R. Gleitman, and Henry Gleitman. “What Some Concepts Might Not Be.” In Concepts: Core Readings, edited by Eric Margolis and
Stephen Laurence, 225–260. Cambridge, MA: MIT Press, 1999.
Avram, Henriette. The MARC Pilot Project. Final Report. Washington, DC: Library of
Congress, 1968.
Baca, Murtha, and Sherman Clarke. “FRBR and Works of Art, Architecture, and Material Culture.” In Understanding FRBR: What It Is and How It Will Affect Our
Retrieval Tools, edited by Arlene Taylor, 103–10. Westport, CT: Libraries Unlimited, 2007.
Baker, Thomas, Karen Coyle, and Sean Petiya. “Multi-Entity Models of Resource
Description in the Semantic Web.” Library Hi Tech 32, no. 4 (2014): 562–82.
doi:10.1108/LHT-08–2014-0081.
Bean, Carol. Relationships in the Organization of Knowledge. Dordrecht; Boston; Norwell MA: Kluwer Academic Publishers, 2001.
Bekiari, Chryssoula, Martin Doerr, and Patrick Le Boeuf. FRBR: Object-Oriented Definition and Mapping to FRBR(ER), V.1.0.2. International Working Group on
FRBR and CIDOC CRM Harmonisation, 2012.
161
©Karen Coyle CC-BY
162 / BIBLIOGRAPHY
Bennett, Rick, Brian F. Lavoie, and Edward T. O’Neill. “The Concept of a Work
in WorldCat: An Application of FRBR.” Library Collections, Acquisitions,
and Technical Services 27, no. 1 (March 2003): 45–59. doi:10.1016/
S1464-9055(02)00306-8.
Berg, Einar Silset. “A Comparison of Two Relational Models: IFLA’s FRBR Model
and Taniguchi’s Expression-Prioritized Model.” Master’s thesis, Oslo University,
2004.
Berners-Lee, Tim, and Ora Lassila. “The Semantic Web.” Scientific American (May
2001). http://www.scientificamerican.com/article.cfm?id=the-semantic-web.
Bianchini, Carlo, and Mauro Guerrini. “From Bibliographic Models to Cataloging
Rules: Remarks on FRBR, ICP, ISBD, and RDA and the Relationships between
Them.” Cataloging and Classification Quarterly 47, no. 2 (January 2009): 105–
24. doi:10.1080/01639370802561674.
Buckland, Michael K. “What Is a ‘Document’?” Journal of the American Society for
Information Science 48, no. 9 (1997): 804–9.
Calhoun, Karen. The Changing Nature of the Catalog and Its Integration with Other
Discovery Tools. Library of Congress, 2006. http://www.loc.gov/catdir/
calhoun-report-final.pdf.
Carlyle, Allyson. “Understanding FRBR as a Conceptual Model: FRBR and the Bibliographic Universe.” Library Resources and Technical Services 50, no. 4 (2006):
264–73.
Carpenter, Michael, and Elaine Svenonius, eds. Foundations of Descriptive Cataloging
Littleton, CO: Libraries Unlimited, 1985.
Casson, Lionel. Libraries in the Ancient World. New Haven, CT: Yale University Press,
2001.
Chen, Peter Pin-shan. “The Entity-Relationship Model—Toward a Unified View of
Data.” ACM Transactions on Database Systems 1, no. 1 (1976): 9–36.
Cossham, Amanda F. “Bibliographic Records in an Online Environment.” IR Information Research 18, no. 3 (September 2013). http://www.informationr.net/
ir/18–3/colis/paperC42.html.
Coyle, Karen. “FRBR, the Domain Model.” Library Technology Reports 46, no. 2
(2010): 20–25.
.“FRBR, Twenty Years On.” Cataloging and Classification Quarterly 32, no. 4
(December 2014): 1–21. doi:10.1080/01639374.2014.943446.
Cutter, Charles Ammi. “The New Catalogue of the Harvard College Library.” North
American Review (January 1869).
.Rules for a Dictionary Catalog. Washington, DC: US Bureau of Education,
1904.
©Karen Coyle CC-BY
BIBLIOGRAPHY / 163
.Rules for a Dictionary Catalogue. Washington, DC: Library of Congress, 1875.
Davis, Ian, and Richard Newman. Expression of Core FRBR Concepts in RDF. 2005.
http://vocab.org/frbr/core.html.
Delsey, Tom. RDA Database Implementation Scenarios, 2009. http://www.rda-jsc
.org/docs/5editor2rev.pdf.
.“Standards for Descriptive Cataloging: Two Perspectives on the Past Twenty
Years.” In The Conceptual Foundations of Descriptive Cataloging, edited by Elaine
Svenonius, 51–60. San Diego, CA: Academic Press, 1989.
Denton, William. “FRBR and the History of Cataloging.” In Understanding FRBR,
35–57, n.d.
.“My Speaking Notes for ‘What We Talk About When We Talk About FRBR,’“
Presentation at Code4Lib 2009” 2009, Providence, RI, February 2009.
http://yorkspace.library.yorku.ca/xmlui/bitstream/handle/10315/2528/frbr
_code4lib09_denton_notes.pdf.
Dervin, B. “The Everyday Information Needs of the Average Citizen: A Taxonomy for
Analysis.” In Information for the Community, edited by Manfred Kochen and
Joseph C. Donohue. Chicago: American Library Association, 1976.
Dickey, Timothy J. “FRBRization of a Library Catalog: Better Collocation of Records,
Leading to Enhanced Search, Retrieval, and Display.” Information Technology
and Libraries 27 (2008): 23−32. doi:10.6017/ital.v27i1.3260.
Doerr, Martin, and Patrick LeBoeuf. “Modelling Intellectual Processes: The FRBRCRM Harmonization” In Digital Libraries: Research and Development. First
International DELOS Conference, Pisa, Italy, February 13-14, 2007. Berlin Heidelberg, Springer-Verlag, 2007. doi:10.1007/978-3-540-77088-6_11.
Doerr, Martin, Guillaume Boutard, Günther Görz, Dolores Iorizzo, Max Jacob, Carlos
Lamsfus, Mika Nyman, et al. FRBR Object-Oriented Definition and Mapping to
FRBRer, 2007.
Domanovszky, Ákos. Functions and Objects of Author and Title Cataloguing: A Contribution to Cataloguing Theory. München: Verlag Dokumentation. 1975.
Dunsire, Gordon. “Representing the FR Family in the Semantic Web.” Cataloging and
Classification Quarterly 50, no. 5–7 (June 2012): 724–41. doi:10.1080/
01639374.2012.679881.
.Resource and Work, Expression, Manifestation, Item, 2013. http://www.ifla.org/
files/assets/cataloguing/isbd/OtherDocumentation/resource-wemi.pdf.
Garrison, Dee. Apostles of Culture: The Public Librarian and American Society, 1876–
1920. New York: Macmillan Information, 1979.
Genette, Gerard. Paratexts: Thresholds of Interpretation. Cambridge [Eng.]: Cambridge
University Press, 1997.
©Karen Coyle CC-BY
164 / BIBLIOGRAPHY
Gorman, Michael. “AACR3? Not!” ALCTS Papers on Library Technical Services and
Collections (1998): 19–29.
Gorman, Michael, and Paul W. Winkler, eds. “25.1 Use of Uniform Titles.” In
Anglo-American Cataloguing Rules; Second Edition, 441–42. Chicago: American
Library Association, 1978.
Gorman, Michael, Paul W. Winkler, and American Library Association. Anglo-American Cataloguing Rules. 2nd ed. London: Library Association, 1978.
Hardesty, Juliet L., Steven Harris, Anna Coogan, and Mark Notess. Scherzo Usability
Test Report: Testing a FRBR Search Interface for Music. Indiana University; Variations project, 2012.
Hickey, Thomas B., and Edward T. O’Neill. “Experiments with the IFLA Functional
Requirements for Bibliographic Records (FRBR),” 2002, 1–15.
Hickey, Thomas B., and Jenny Toves. “FRBR Work-Set Algorithm.” Dublin, OH:
OCLC, 2005. http://www.oclc.org/content/dam/research/activities/
frbralgorithm/2005-04.pdf.
Hillmann, Diane, Karen Coyle, John Phipps, and Gordon Dunsire. “RDA Vocabularies: Process, Outcome, Use.” D-Lib Magazine 16, no. 1/2 (2010).
doi:10.1045/january2010-hillmann.
Howarth, Lynne C. “FRBR and Linked Data: Connecting FRBR and Linked Data.”
Cataloging and Classification Quarterly 50, no. 5–7 (June 2012): 763–76.
doi:10.1080/01639374.2012.680835.
IFLA (International Federation of Library Associations). Statement of International
Cataloging Principles, (2009).
.Statement of Principles Adopted by the International Conference on Cataloguing
Principles. Paris, 1961.
IFLA Cataloguing Section. International Standard Bibliographic Description (ISBD),
2011. http://www.ifla.org/files/assets/cataloguing/isbd/isbd-cons_20110321
.pdf.
IFLA Cataloguing Section. ISBD Review Group. Mapping ISBD Elements to FRBR
Entity Attributes and Relationships. IFLA, 2004. http://www.ifla.org/files/
assets/cataloguing/isbd/isbd-frbr-mapping.pdf.
IFLA Study Group on the Functional Requirements for Bibliographic Records. Functional Requirements for Bibliographic Records, 2009. http://archive.ifla.org/
VII/s13/frbr/frbr_2008.pdf.
IFLA Standing Committee of the Section on Cataloguing, Terms of Reference for a
Study of the Functional Requirements for Bibliographic Records, 1992.
©Karen Coyle CC-BY
BIBLIOGRAPHY / 165
International Working Group on FRBR and CIDOC CRM Harmonisation. FRBR
Object-Oriented Definition and Mapping from FRBRER, FRAD and FRSAD.
CIDOC, 2015. http://www.cidoc-crm.org/docs/frbr_oo//frbr_docs/
FRBRoo_V2.1_2015February.pdf.
Jones, Ed. “The FRBR Model as Applied to Continuing Resources.” Library Resources
and Technical Services 49, no. 4 (October 2005.): 227–42.
Jones, Ed, Caitlin Hunter, and Jenn Riley. FRBR, FRAD and Music: Theory and
Practice Presentation, Music Library Association, 2010. http://bcc
.musiclibraryassoc.org/Presentations/FRBR-FRAD-MLA-2010.pdf.
Jonsson, Gunilla. “Cataloguing of Hand Press Materials and the Concept of Expression
in FRBR,” Cataloging and Classification Quarterly, November 2013 (2009):
37–41. doi:10.1300/J104v39n03.
Kalf, Ruth. “FRBR: An Opportunity for Map Collections and Map Users?” LIBER
Quarterly 18, no. 2 (September 2008). https://liber.library.uu.nl/index.php/
lq/article/viewFile/7927/8185.
Knapp, John F. “Design Considerations for the MARC Magnetic Tape Formats.”
Library Resources and Technical Services 12, no. 3 (1968): 275–85.
Kochen, Manfred, and Joseph C. Donohue, eds. Information for the Community.
Chicago: American Library Association, 1976.
Krajewski, Markus. Paper Machines: About Cards and Catalogs, 1548−1929. Cambridge,
MA: MIT Press, 2011.
Krier, Laura. “Serials, FRBR, and Library Linked Data: A Way Forward.” Journal of
Library Metadata 12, no. 2–3 (April 2012): 177–87. doi:10.1080/19386389
.2012.699834.
Lavoie, Brian F., and Edward T. O’Neill. “The Concept of a Work in WorldCat: An
Application of FRBR.” Library Collections, Acquisitions, and Technical Services
27, no. 1 (2003): 32.
Le Boeuf, Patrick. “Foreword.” Cataloging and Classification Quarterly 50, no. 5–7
(June 2012): 355–59. doi:10.1080/01639374.2012.682001.
.“FRBR: Hype or Cure-All? Introduction.” In Functional Requirements for Bibliographic Records: Hype or Cure-All? 1–13. New York: Haworth Information
Press, 2005.
.Functional Requirements for Bibliographic Records (FRBR): Hype or Cure-All?
Edited by Patrick Le Boeuf. New York: Haworth Information Press, 2005.
Le Boeuf, Patrick, Barbara Tillett, and Elena Escolano Rodriguez. Brave New FRBR
World. Cairo: IME ICC 3, 2005.
©Karen Coyle CC-BY
166 / BIBLIOGRAPHY
Lerner, Frederick Andrew. The Story of Libraries: From the Invention of Writing to the
Computer Age. New York: Continuum, 2001.
Library of Congress Working Group on the Future of Bibliographic Control. On the
Record. Washington, DC: Library of Congress, 2008.
Lubetzky, Seymour. Seymour Lubtezky: Writings on the Classical Art of Cataloging.
Edited by Elaine Svenonius and Dorothy McGarry. Englewood, CO, Libraries
Unlimited, 2001.
Lugg, Rick, and R2 Consulting LLC. “Working Group on the Future of Bibliographic
Control: Economics and Organization of Bibliographic Data Economics of Bibliographic Control,” 2007. http://www.loc.gov/bibliographic-future/
meetings/docs/ricklugg-july9-2007.pdf.
Madison, Olivia M. A. “The IFLA Functional Requirements for Bibliographic Records:
International Standards for Universal Bibliographic Control.” Library Resources
and Technical Services 44, no. 3 (July 2000): 153–159.
.“The Origins of the IFLA Study on Functional Requirements for Bibliographic
Records.” Cataloging and Classification Quarterly 39, no. 3–4 (2005): 15–37.
.“Standards for International Bibliographic Control: Proposed Basic Data
Requirements for the National Bibliographic Record.” In Universal Bibliographic
Control and International MARC Core Programme (UBCIM), 1–11. Copenhagen: ICNBS, 1998.
Malssen, Kara Van, Caitlin Hunter, and Andrea Leigh. BIBFRAME AV Modeling
Study : Defining a Flexible Model for Description of Audiovisual Resources.
Library of Congress, 2014. http://www.loc.gov/bibframe/pdf/bibframe
-avmodelingstudy-may15-2014.pdf.
Manzanos, Norberto. “Item, Document, Carrier: An Object Oriented Approach.” Cataloging and Classification Quarterly 50, no. February 2014 (2012): 439–55. doi:
10.1080/01639374.2012.681274.
Marcum, Deanna B. “Response to On the Record : Report of the Library of Congress
Working Group on the Future of Bibliographic Control.” Washington, DC: Library
of Congress. 2008. http://www.loc.gov/bibliographic-future/news/
LCWGResponse-Marcum-Final-061008.pdf.
Margolis, Eric, and Stephen Laurence. Concepts: Core Reading. Cambridge, MA: MIT
Press, 1999.
Maunsell, Andrew. The Catalogue of English Printed Books. Farnborough, UK: Gregg,
Press, 1965. First published 1595.
Maxwell, Robert L. FRBR: A Guide for the Perplexed. Chicago: American Library Association, 2008.
©Karen Coyle CC-BY
BIBLIOGRAPHY / 167
Miksa, Shawne D. “Understanding Support of FRBR’s Four User Tasks in MARCEncoded Bibliographic Records.” Bulletin of the American Society for Information Science and Technology, August/September 2007: 24–26.
Miller, David, and Patrick Le Boeuf. “‘Such Stuff as Dreams Are Made On’: How Does
FRBR Fit Performing Arts ?” Cataloging and Classification Quarterly 39, no.
3–4 (2005): 151–78. doi:10.1300/J104v39n03_10.
Murphy, Gregory L. The Big Book of Concepts. Cambridge, MA: MIT Press, 2004.
Murray, Ronald J. The FRBR-Theoretic Library : The Role of Conceptual Data Modeling in
Cultural Heritage Information System Design. Washington, DC: Library of Congress, 2008. http://www.bl.uk/ipres2008/presentations_day1/26_Murray.pdf.
Murray, Ronald J., and Barbara B. Tillett. “From Moby-Dick to Mashups : Thinking
about Bibliographic Networks Where This Is Going :” In American Library Association 2010 Annual Conference. Washington, DC: Library of Congress, 2010.
http://www.slideshare.net/RonMurray/from-mobydick-to-mashups.
Noerr, Peter, Paula Goossens, Dan Matei, Petra Otten, and Susanna Peruginelli. “User
Benefits from a New Bibliographic Model: Follow-up of the IFLA Functional
Requirements Study.” In 64th IFLA General Conference, 1–6. IFLA, 1998.
http://archive.ifla.org/IV/ifla64/084-126e.htm.
O’Neill, Ed, Anders Cato, Paula Goossens, Judy Kuhagen, Barbara Tillett, Carol Van
Nuys, and Maja Žumer. Final Report of the Working Group on Aggregates, 2011.
http://www.ifla.org/files/assets/cataloguing/frbrrg/AggregatesFinalReport.pdf.
O’Neill, Edward T. “FRBR : Functional Requirements for Bibliographic Records
Application of the Entity-Relationship Model to Humphry Clinker.” Library
Resources and Technical Services 46, no. 4 (2002): 150–59.
O’Neill, Edward T., and Maja Žumer. Aggregates as Manifestations. IFLA Working
Group on Aggregates, n.d.
Pearl, Nancy. Book Lust: Recommended Reading for Every Mood, Moment, and Reason.
Seattle: Sasquatch Books. 2003.
Peponakis, Manolis. “Conceptualizations of the Cataloging Object: A Critique on
Current Perceptions of FRBR Group 1 Entities.” Cataloging and Classification
Quarterly 50, no. 5–7 (June 2012): 587–602. doi:10.1080/01639374.2012
.681275.
.“Libraries’ Metadata as Data in the Era of the Semantic Web Manolis Peponakis.” Journal of Library Metadata 13, no. 4 (2013): 330–48. doi:10.1080/
19386389.2013.846618.
Perkinson, Richard C. Data Analysis: The Key to Data Base Design. Wellesley, MA:
QED Information Sciences, 1984.
©Karen Coyle CC-BY
168 / BIBLIOGRAPHY
Petrucciani, Alberto. “From the FRBR Model to the Italian Cataloging Code (and
Vice Versa?).” Cataloging and Classification Quarterly 50, no. 5–7 (June 2012):
603–21. doi:10.1080/01639374.2012.680841.
Picco, Paola, and Virginia Ortiz Repiso. “The Contribution of FRBR to the Identification of Bibliographic Relationships: The New RDA-Based Ways of Representing
Relationships in Catalogs.” Cataloging and Classification Quarterly 50, no. 5–7
(June 2012): 622–40. doi:10.1080/01639374.2012.680847.
Pisanski, Jan, and Maja Žumer. “Functional Requirements for Bibliographic Records:
An Investigation of Two Prototypes.” Program: Electronic Library and Information Systems 41, no. 4 (2007): 400–417. doi:10.1108/00330330710831602.
.“Mental Models of the Bibliographic Universe. Part 1: Mental Models of
Descriptions.” Journal of Documentation 66, no. 5 (2010): 643–67.
.“Mental Models of the Bibliographic Universe. Part 2: Comparison Task and
Conclusions.” Journal of Documentation 66, no. 5 (2010): 668–80.
Ranganathan, S. R. The Five Laws of Library Science. Bombay, New York: Asia Pub.
House, 1963.
Ranganathan, S. R., and M. A. Gopinath. Prolegomena to Library Classification. New
Delhi: Ess Ess Publications, 1989.
Rather, Lucia J. “Special Characters and Diacritical Marks Used in Roman Alphabets.”
Library Resources and Technical Services 12, no. 3 (1968): 285–95.
Renear, Allen, and Yunseon Choi. “Some Conceptual Modeling Issues in FRBR,”
2006, 1–7.
Renear, Allen H., and David Dubin. “Three of the Four FRBR Group 1 Entity
Types Are Roles, Not Types.” Proceedings of the American Society for Information Science and Technology 44, no. 1 (October 2008): 1–19. doi:10.1002/
meet.1450440248.
Renear, Allen H., and Yunseon Choi. “Modeling Our Understanding, Understanding
Our Models—The Case of Inheritance in FRBR.” Proceedings of the American Society for Information Science and Technology. New York: Wiley, 2006
doi:10.1002/meet.14504301179.
Riley, Jenn. “Application of the Functional Requirements for Bibliographic Records
(FRBR) to Music.” In Proceedings of the International Society for Music Information Retrieval 2008, 439–44, 2008.
.“Enhancing Interoperability of FRBR-Based Metadata.” In International Conference on Dublin Core and Metadata Applications, 31–43, 2010.
Riley, Jenn, Caitlin Hunter, Chris Colvard, and Alex Berry. Definition of a FRBR-Based
Metadata Model for the Indiana University Variations3 Project, 2007.
http://dlib.indiana.edu/projects/variations3/docs/v3FRBRreport.pdf.
©Karen Coyle CC-BY
BIBLIOGRAPHY / 169
Riva, Pat, and Chris Oliver. “Evaluation of RDA as an Implementation of FRBR and
FRAD.” Cataloging and Classification Quarterly 50, no. 5–7 (June 2012): 564–
86. doi:10.1080/01639374.2012.680848.
Rose, Mary Z. “The Ship Has Sailed and We Aren’t on It: How Catalogers Could Support User Tasks and Why We Won’t.” Journal of Library Metadata 12, no. 2–3
(April 2012): 127–39. doi:10.1080/19386389.2012.699828.
Rust, Godfrey, and Mark Bide. The Indecs Metadata Framework; Principles, Model and
Data Dictionary. Indecs Framework Ltd, 2000.
Schaub, Jacob, Yin Zhang, and Athena Salaba. “What Is Next for Functional Requirements for Bibliographic Records? A Delphi Study.” Library Quarterly 79, no. 2
(2009): 233–55.
Schmidt, Raymond. “Composing in Real Time: Jazz Performances as ‘Works’ in the
FRBR Model.” Cataloging and Classification Quarterly 50, no. 5–7 (June
2012): 653–69. doi:10.1080/01639374.2012.681601.
Seikel, Michele. “General Notes in Catalog Records versus FRBR User Tasks.” Cataloging and Classification Quarterly 51, no. 4 (May 2013): 420–27. doi:10.1080
/01639374.2012.749318.
Serrai, Alfredo, and Fiammetta Sabba. 2005. Profilo di storia della bibliografia. Milano:
Sylvestre Bonnard.
Shadle, Steve. “FRBR and Serials: An Overview and Analysis.” Serials Librarian 50,
no. 1–2 (2008): 83–103. doi:10.1300/J123v50n01_09.
Shotton, David. Extending FRBR within FaBiO, 2011. http://opencitations
.wordpress.com/2011/06/29/extending-frbr-within-fabio/.
Smiraglia, Richard P. “Be Careful What You Wish For: FRBR, Some Lacunae, a
Review.” Cataloging and Classification Quarterly 50, no. 5–7 (June 2012):
360–68. doi:10.1080/01639374.2012.682254.
.“Further Reflections on the Nature of ‘A Work’: An Introduction.” Cataloging
and Classification Quarterly 33, no. 3–4 (August 2002): 1–11. doi:10.1300/
J104v33n03_01.
.“Instantiation: Toward a Theory.” In Data, Information and Knowledge in a
Networked World; Annual Conference of the Canadian Association for Information Science. London, Ontario. June 2–4 2005, 1–8, 2005.
.“A Meta-Analysis of Instantiation as a Phenomenon of Information Objects.”
Culture Del Testo e Del Documento 9, no. 25 (2008): 5–25.
.“Musical Works as Information Retrieval Entities: Epistemological Perspectives,”
2nd Annual International Symposium on Music Information Retrieval, Bloomington, 2001. http://ismir2001.ismir.net/pdf/smiraglia.pdf.
.The Nature of “A Work”: Implications for the Organization of Knowledge.
Lanham: Scarecrow Press, 2001.
©Karen Coyle CC-BY
170 / BIBLIOGRAPHY
.Works as Entities for Information Retrieval. Binghampton, NY: Haworth Information Press. 2002.
Spalding, Tim. IMPORTANT: LibraryThing Dives into Editions and Expressions,
2011. http://www.librarything.com/topic/109523.
Svenonius, Elaine. The Conceptual Foundations of Descriptive Cataloging. San Diego,
CA: Academic Press, 1989.
Taniguchi, Shoichi. “A Conceptual Model Giving Primacy to Text-Level Bibliographic
Entity in Cataloging: A Detailed Discussion,” Journal of Documentation 58, no.
4 (4) (2003): 363−82.
.“Expression-Level Bibliographic Entity Records: A Trial on Creation from
Pre-Existing MARC Records,” (October 2008): 37–41. doi:10.1300/
J104v38n02_05.
.“User Tasks in the RDA-Based Model.” Cataloging and Classification Quarterly
51, no. 7 (October 2013): 788–815. doi:10.1080/01639374.2013.800623.
.“Viewing RDA from FRBR and FRAD: Does RDA Represent a Different Conceptual Model?” Cataloging and Classification Quarterly 50, no. 8 (November
2012): 929–43. doi:10.1080/01639374.2012.712631.
Taylor, Arlene, ed. Understanding FRBR: What It Is and How It Will Affect Our
Retrieval Tools. Westport, CT: Libraries Unlimited, 2007.
Thomale, Jason. “Interpreting MARC: Where’s the Bibliographic Data?” Code41ib
Journal, no. 11 (September 21, 2010).
Tillett, Barbara B. “AACR3: Resource Description and Access.” Presentation to Library
of Congress, 2004.
.“Bibliographic Relationships.” In Relationships in the Organization of Knowledge,
19–36. Dordrecht: Kluwer, 2001.
.“Bibliographic Relationships: Toward a Conceptual Structure of Bibliographic
information Used in Cataloging.” Dissertation, University of California, Los
Angeles, 1988.
.Definition of Aggregates as Works: Tillett Proposal, IFLA Working Group on
Aggregates, 2007.
.“FRBR and Cataloging for the Future.” Cataloging and Classification Quarterly
39, no. 3 (2005): 197–205.
.“IFLA Study on the Functional Requirements of Bibliographic Records:
Theoretical and Practical Foundations,” Proceedings of the 60th IFLA General
Conference, August 21−27, 1994.
.What Is FRBR? A Conceptual Model for the Bibliographic Universe. Washington,
DC: Library of Congress. Cataloging Distribution Service, 2003.
©Karen Coyle CC-BY
BIBLIOGRAPHY / 171
Turner, James M, and Abby A Goodrum. “Modeling Videos as Works.” Cataloging and Classification Quarterly 33, no. 3–4 (2009): 27–38. doi:10.1300/
J104v33n03_03.
University of Minnesota Libraries. A Multi-Dimensional Framework for Academic Support: A Final Report, 2006.
Unsworth, John. “Scholarly Primitives: What Methods Do Humanities Researchers
Have in Common and How Might Our Tools Reflect This?” In Humanities
Computing, Formal Methods, Experimental Practice, Kings College, London, May
13, 2000, 2000.
Verona, Eva. “Literary Unit versus Bibliographic Unit.” In Foundations of Descriptive
Cataloging, ed. By Michael Carpenter and Elaine Svenonius, 155–175. Littleton,
CO: Libraries Unlimited, 1985.
Weiss, Paul J., and Steve Shadle. “FRBR in the Real World.” Serials Librarian 52, no.
1–2 (2007): 93–104. doi: 10.1300/J123v52n01_09.
Wiesenmüller, Heidrun. Some Comments on the Final Report of the FRBR Working
Group on Aggregates. Stuttgart: Stuttgart Media University, 2011.
https://www.mendeley.com/download/public/5179751/4486442993/
59a97ce9df66ece320e2dc2e5e6421730c0531d4/dl.pdf.
Willer, Mirna, and Ana Barbari. “FRBR/FRAD and Eva Verona’s Cataloging Code:
Toward the Future Development of Croatian Cataloging Code.” Cataloging and
Classification Quarterly 50, no. 5–7 (June 2012): 542–63. doi:10.1080/
01639374.2012.681613.
Wilson, Patrick. “Interpreting the Second Objective of the Catalog.” Library Quarterly
59, no. 4 (1989): 339–53.
.“The Second Objective.” In The Conceptual Foundations of Descriptive Cataloging, edited by Elaine Svenonius, 5–16. San Diego, CA: Academic Press, 1989.
.Two Kinds of Power: An Essay on Bibliographical Control. University of California
Publications in Librarianship. Berkeley, CA: University of California Press, 1968.
Wilson, T. D. “Information-Seeking Behaviour and Information Use.” Social Science
Information Studies 4 (1984): 197–204.
.“Models in Information Behavior Research.” Journal of Documentation 22, no. 3
(1966): 266–68. doi:10.1108/eb026404.
Yee, Martha M.Cataloging Rules and the RDF Model. 2007. http://myee.bol.ucla.
edu/catrul&rdfIntro.htm.
.“FRBR and Moving Image Materials: Content (Work and Expression) Versus
Carrier (Manifestation).” In Understanding FRBR: What It Is and How It Will
Affect Our Retrieval Tools, edited by Arlene Taylor, 117–30. Westport, CT:
Libraries Unlimited, 2007.
©Karen Coyle CC-BY
172 / BIBLIOGRAPHY
.“FRBRization: A Method for Turning Online Public Finding Lists into Online
Public Catalogs.” Information Technology and Libraries 24, no. 3 (2005): 77–95.
Zhang, Yin, and Athena Salaba. “What Do Users Tell Us About FRBR-Based Catalogs?” Cataloging and Classification Quarterly 50, no. 5–7 (June 2012): 705–23.
doi:10.1080/01639374.2012.682000.
Zipf, George Kingsley. Human Behavior and the Principle of Least Effort: An Introduction to Human Ecology. Cambridge, MA: Addison-Wesley Press. 1949.
©Karen Coyle CC-BY
INDEX
A
AACR (Anglo-American Cataloguing Rules),
11, 22
AACR2 (Anglo-American Cataloguing Rules,
Second Edition), 24
Abbot, Ezra, 49
accompanying relationships, 21
aggregates, 90, 130–135
AI (artificial intelligence), 54, 59
A.L.A. Cataloging Rules, 11, 22
alternative models using FRBR concepts
FaBiO (FRBR-aligned Bibliographic
Ontology), 154–156
FRBRcore, 148–150
FRBRoo, 150–153
<indecs> event-oriented model, 153–154
overview, 147–148
American Library Association, 49
Antelman, Kristin, 119
Apostles of Culture (Garrison), 44
art and artwork cataloging, 122–123
Avram, Henriette, 50, 72
alternative models using FRBR concepts,
147–156
entity relationship modeling and Semantic
Web models, 138–142
FRBR in RDF, 142–147
overview, 137–138
bibliographic families and works, 13–15, 16,
87, 110
bibliographic function of library catalog,
39–40
bibliographic relationships, 20–22, 92–93
Bibliotheca Universalis (Gessner), 8
The Big Book of Concepts (Murphy), 7
Bobst Library (New York University), 16
books versus works, 12
British Museum Library, 9
Burke Theological Library, 16
C
Calhoun, Karen, 66
card catalogs
as data models, 30
out-of-date, 51
overview, 49–50
cardinality, 58, 59
Carlyle, Allyson, 98
cataloging
efficiency, 113–117
rules, Lubetzky’s analysis of, 11
works in cataloging practice, 22–27
Cataloging Cultural Objects (CCO), 122–123
Choi, Yunseon, 126
B
Baca, Murtha, 123
Basic Level National Bibliographic Record
(BLNB), 104–106
basic level of categorization, 17–19
Berners-Lee, Tim, 54
BIBFRAME, 6, 85, 145–147
bibliographic context, works in, 10–20
bibliographic description and the Semantic
Web
173
©Karen Coyle CC-BY
174 / INDEX
Clarke, Sherman, 123
classes, 61–62
classification systems, 6. See also specific
classification systems
cognitive view of works, 17–19
collective titles, 26
collocation, 23, 112–113
Colon Classification, 6
“Composing in Real Time: Jazz Performances
as ‘Works’ in the FRBR Model”
(Schmidt), 120
computerized catalogs, 50–53
concept, object, event, and place
and CCO, 123
overview, 83, 91–92
conceptual model
FRBR as, 83–86, 157–158
overview, 33
Conceptual Reference Model of the
International Council of Museums
(CIDOC CRM), 150
corporate bodies. See persons, corporate bodies,
families
Coyle’s cognitive view of works, 17–19
creators, 4–5
Cutter, Charles Ammi, 5, 9, 35–38, 39, 42,
45, 48, 51, 102
D
Data Analysis: The Key to Data Base Design
(Perkinson), 73
data models
history of, 30–34
library data models, 34–46
overview, 29–30, 81
data sharing
overview, 117–118
on Semantic Web, 141–142
database management systems, 31, 51–53
database models in RDA, 140–141
Davis, Ian, 148–149
degrees of belonging, 7
Delsey, Tom, 72, 79
derivative relationships, 21
descriptive relationships, 21
development and review of FRBR, 69–74
Dewey, Melvil, 5, 49, 50
Dewey Decimal Classification, 6, 49
digital versions of resources, 116–117
disjoints, 127–128, 129–130, 143
Doerr, Martin, 152
Domanovszky, Ákos, 22
Dublin Core (DC), 60–61
Dublin Core Metadata Community, 144
Dublin Core Metadata Initiative (DCMI), 60
Dublin Core standard for metadata, 60–61,
144
Dulabahn, Elizabeth, 72
Dunsire, Gordon, 82
E
economic reasons for FRBR, 71–72
editions versus works, 8–10
endeavor class, 149
entity-relationship (E-R) model
overview, 33–34, 66–67, 75–80
and Semantic Web models, 138–142
of work, expression, manifestation, and item
(WEMI), 94–95
equivalence, 58
equivalence relationships, 21
Expansive Classification, 5
expression-dominant model for works, 19–20,
110, 155
expressions
overview, 88–89
users seeking items at level of, 110–111
F
FaBiO (FRBR-aligned Bibliographic
Ontology), 154–156
families. See persons, corporate bodies, families
film cataloging, 121–122
flat file technology, 30–31
format neutrality, 118–124
Foucault, Michel, 5
FRAD (Functional Requirements for
Authority Data), 91, 101
©Karen Coyle CC-BY
INDEX / 175
FRBR: A Guide for the Perplexed (Maxwell),
130
FRBR Final Report, 66, 71–73, 77, 84,
85–90, 93, 99–101, 104–105, 106–108,
124, 129–130, 131–132, 157–158
FRBR (Functional Requirements for
Bibliographic Records). See also
alternative models using FRBR concepts;
objectives for FRBR
as conceptual model, 83–86
overview, 6, 66–68
FRBR in RDF
BIBFRAME, 145–147
FRBRer, 142–143
RDA in RDF, 144–145
“FRBR in the Real World” (Weiss and
Shadle), 119
FRBR Review Group, 82, 142
FRBR Study Group, 54, 65, 66–67, 70–73,
76–78, 82, 84, 85, 99, 107, 124, 132
FRBRcore, 148–150
FRBRer, 142–143
FRBRoo, 150–153
“The FRBR Model as Applied to Continuing
Resources” (Jones), 120
FRSAD (Functional Requirements for Subject
Authority Data), 101
functional requirements, 99–104, 157–158
The Functions and Objects of Author and Title
Cataloguing (Domanovszky), 22
The Future of the Descriptive Cataloging Rules
(Gorman), 98
G
Galbraith, Robert, 5
GARE (Guidelines for Authority and
Reference Entries), 105
Garrison, Dee, 44
Genette, Gerard, 131
Georgetown University Library, 16
Gessner, Conrad, 8–9
GoodReads, 145
Goodrum, Abby A., 122
Gorman, Michael, 98
graphs, 55–56
Group 1. See work, expression, manifestation,
and item
Group 2. See persons, corporate bodies,
families
Group 3. See concept, object, event, and
place
groupings of entities, 82–83
H
Hendler, James, 54
hierarchy and inheritance, 125–128
I
identity management, 4–5
IFLA (International Federation of Library
Associations and Institutions), 19, 69
ILS (integrated library system), 41
<indecs> event-oriented model, 10, 153–154
inferencing, 58
inheritance and hierarchy, 125–128
International Cataloguing Principles, 22, 42, 44,
69, 103
International Conference on Cataloguing
Principles (Paris Principles), 36–38, 45
International Congress on National
Bibliographic Services, 70
International Society for Music Information
Retrieval, 121
Internet Movie Database (IMDB), 145
Internet search engines, 40–41
“Interpreting the Second Objective of the
Catalog” (Wilson), 15, 28
ISBD (International Standards for
Bibliographic Description), 69, 70–71,
104–105
ISBN (International Standard Book Number),
102
items, 90
J
Joint Steering Committee (JSC), 140–141,
144
Jones, Ed, 120
©Karen Coyle CC-BY
176 / INDEX
K
Kalf, Ruth, 124
Krajewski, Marcus, 30, 49
Krier, Laura, 120
L
Lassila, Ora, 54
Lavoie, Brian, 114, 115
Le Boeuf, Patrick, 106, 152
Library Bureau, 49
library catalog
bibliographic function of, 39–40
within entire library workflow, 41–42
goals of, 35–38, 42–45
overview, 34–35
and users, 42–45
library catalog technology
card catalogs, 49–50
computerized catalogs, 50–53
database management systems, 51–53
historical overview, 48–53
MARC records, 50
online catalogs, 50–53
present and future of, 53–62
printed book catalogs, 48–49
library data models, 34–46
Library of Congress, 45, 50
Library of Congress Classification, 6
library workflow, library catalog within entire,
41–42
LibraryThing, 112, 145
linked data, 57–58, 62
Linked Open Vocabularies project, 150
logical models, 33
Lubetzky, Seymour, 4, 10–13, 18, 19, 22, 25,
28, 45, 70, 99
Lubetzky’s work view, 10–13
M
Madison, Olivia, 70, 73, 74, 107
manifestations
aggregates as, 132
attributes, 100–103
overview, 89–90
many-to-many relationships, 76, 128, 133
map cataloging, 124
MARC (Machine Readable Cataloging), 50,
51, 53, 78–80
MARC 21 standard for metadata, 61, 106,
146
The MARC Pilot Project (Avram), 50
MARCXML, 53
Maxwell, Robert, 92, 130
McEathron, Scott R., 124
“Mental Models of the Bibliographic
Universe” (Pisanski and Žumer), 109
metadata
Dublin Core standard, 60–61, 144
MARC 21 standard, 61, 106, 146
“Modeling Our Understanding,
Understanding Our Models” (Renear
and Choi), 126
moving images cataloging, 121–122
Mumford, L. Quincy, 50
Murphy, Gregory L., 7
music cataloging, 120–121
music uniform titles, 26–27
MusicBrainz, 145
N
national libraries, basic bibliographic record
for, 104–106
The Nature of “A Work”: Implications for the
Organization of Knowledge (Smiraglia),
3, 15
Newman, Richard, 148–149
NISO (National Information Standards
Organization), 42
Non-Unique Name Assumption (NUNA),
57, 59, 60
normalization, 33
O
object-oriented (OO) models, 34, 54,
150–153
objectives for FRBR
cataloging efficiency, 113–117
data sharing, 117–118
©Karen Coyle CC-BY
INDEX / 177
format neutrality, 118–124
functional requirements, 99–104
national level bibliographic record, 104–
106
overview, 97–98
user needs, 106–113
OCLC WorldCat database, 16, 28, 45, 112,
114, 115, 116
Ohio College Library Center, 51
On the Record (Library of Congress), 66
one-to-many relationships, 76
O’Neil, Ed, 132
online catalogs, 50–53
ontology, 58
OPAC (Online Public Access Catalogues), 38
Open World Assumption, 57, 59, 60, 140
OpenLibrary, 112
original work, 110
Otlet, Paul, 5
out-of-date card catalogs, 51
OWL (Web Ontology Language), 58–60,
142, 143
P
Panizzi, Anthony, 9, 37, 38, 65
Panizzi’s rules, 37, 38, 65
Paper Machines (Krajewsky), 49
paratexts, 131–132
Paratexts: Thresholds of Interpretation
(Genette), 131
Paris Principles, 36–38, 45
Pearl, Nancy, 40
Perkinson, Richard C., 73
persons, corporate bodies, families
and CCO, 123
overview, 83, 91
Pettee, Julia, 11
Philosophical Investigations (Wittgenstein), 7
physical models, 33
Pisanski, Jan, 109, 113
Principle of Least Effort, 7
Principles of Cataloging, Phase I (Lubetzky),
12
printed book catalogs, 48–49
R
Ranganathan, S. R., 6, 13
RDA Implementation Database Scenarios, 79
RDA (Resource Description and Access)
database models, 140–141
FRBR implementation in, 66
overview, 6, 144
in RDF, 144–145
standards, development of, 65–68
Tillett’s analysis of relationships and, 22
RDF (Resource Description Framework),
53–54, 55–56, 137, 142–147
reader’s services, 40
relational database management systems
(RDBMS), 31–33, 78, 139–140
Renear, Allen, 126
Research Libraries Group, 51
responsible entity class, 149
review and development of FRBR, 69–74
Riley, Jenn, 121
RLIN system, 51
Rowling, J. K., 5
Rules for a Dictionary Catalog (Cutter), 9–10
S
Scherzo project, 111–112
Schmidt, Raymond, 120
search engines, 40–41
Semantic Publishing and Referencing (SPAR)
suite, 154
Semantic Web. See also bibliographic
description and the Semantic Web
advantages of, 140
classes, 61–62
data sharing, 141–142
disadvantages of, 140
entity-relationship modeling and Semantic
Web models, 138–142
inferencing, 58
linked data, 57–58
Non-Unique Name Assumption, 57, 60
Open World Assumption, 57, 60
overview, 34, 54–56
semiotic view of works, 15–17
©Karen Coyle CC-BY
178 / INDEX
semiotics, 15
sequential relationships, 21
serials cataloging, 119–120
Shadle, Steve, 119
shared characteristics relationships, 21
Smiraglia, Richard, 3, 15–17, 18, 22, 28, 93,
114
Smiraglia’s semiotic view of works, 15–17
standards, development of, 65–68
Stockholm Seminar on Cataloguing, 69
sub-classes, 152, 154
subject access, 5–6
subject class, 149
subjects, 4. See also concept, object, event, and
place
super-classes, 149–150, 152
Svenonius, Elaine, 72
T
Taniguchi, Shoichi, 18, 19–20, 27, 110, 155
Taniguchi’s expression-dominant model for
works, 19–20, 110, 155
taxonomies, 127
technology, library catalog. See library catalog
technology
Terms of Reference (Delsey and Avram),
72–73, 75, 92, 98, 108
Tillett, Barbara, 21–22, 70, 72, 77, 87, 93,
132, 133
triples, 140
Tucker, Ben, 72
Turner, James M., 122
Two Kinds of Power (Wilson), 13, 28, 39–40,
44
U
uniform titles, 22–27, 104
unique material cataloging, 122–123
Universal Decimal Classification (UDC), 5, 6
Universal Modeling Language (UML), 34, 76
University of California library system, 51
University of Minnesota Libraries, 43
Unsworth, John, 43
URL (Uniform Resource Locator), 55
user needs, 42–45, 106–113
user task analysis, 99–104, 108
users seeking items at level of expressions,
110–111
V
Variations Digital Music Library 2 Project, 121
Vellucci, Sherry, 114
Verona, Eva, 10
VTLS, 112
W
Web Ontology Language (OWL), 58–60, 142,
143
Weiss, Paul, 119
What is FRBR? (Tillett), 87
whole-part relationships, 21
Wiesenmüller, Heidrun, 133
Wilson, Patrick, 4, 13–15, 18, 21, 22, 27, 28,
39, 44, 45, 89, 110
Wilson’s bibliographic families and works,
13–15, 16, 87, 110
Wittgenstein, Ludwig, 7
work, expression, manifestation, and item
(WEMI)
aggregates, 130–135
cataloging efficiency and, 113
and CCO, 123
diagrams of, 94–95, 128
disjoint, entities as, 129–130
E-R model, 94–95
inheritance and hierarchy, 125–128
overview, 83, 86
and user view of the bibliographic universe,
109–110
Working Group for Functional Requirements
for Bibliographic Records, 158–159
Working Group on Aggregates, 132–134
works
aggregates as, 132–133
attributes, 102
in bibliographic context, 10–20
versus books, 12
in cataloging practice, 22–27
©Karen Coyle CC-BY
INDEX / 179
Coyle’s cognitive view, 17–19
defined, 6–8
versus editions, 8–10
Lubetzky’s work view, 10–13
overview, 3–4, 87–88
and relationships, 20–22
Smiraglia’s semiotic view, 15–17
Taniguchi’s expression-dominant model,
19–20, 110, 155
Wilson’s bibliographic families and, 13–15,
16, 87, 110
World Wide Web, 54–57, 60
World Wide Web Consortium, 67, 137
WorldCat database, 16, 28, 45, 112, 114, 115,
116
X
XML version of MARC record, 53
Y
Yee, Martha, 112, 121–122
Z
Zipf, George Kingsley, 8
Žumer, Maja, 109, 113, 132
©Karen Coyle CC-BY